The Price of Intelligence Saga:

1. Episode I: Attack of the Code Editor Clones:

   1. AI Code Editors have the same bold promises as the neobank fintechs of yesteryear, but without factoring in unit economics, we’re due for the same harsh realities.

2. Episode II: The Token Economics Strike Back:

   1. The math of compute breaks down when we break down the price of the prompt and the token taxes for AI code editing

3. Episode III: The Hyperscaler Menace:

   1. The unsustainable economics problem goes deeper than the code editors, down to the looming crisis of the AI infrastructure that powers the entire ecosystem

In late 2021-2022, the fintech scene exploded with a surge of neobanks for every affinity and demographic. They quickly amassed users, deposits, and purchase volume to excite any VC. But everyone seemed to forget that with fintech, unlike SaaS, just getting PMF isn't enough because the costs to serve are real and significant. If pricing isn't right, you're just selling a dollar for $0.50 and your users are taking advantage of you.

Cracks appeared in the ecosystem as bank platform fees, processor costs, and compliance expenses caught up to them. Today, we only see the remnants of many sought-after neobanks.

Today, the same pattern is playing out in AI coding tools, but with a greater disconnect between growth and unit economics. Fintech companies assumed they could acquire users for free and monetize them later, and AI code editor companies assumed they could get developers hooked on unprofitable $20/month tools and either rate-limit them into higher tiers or upsell them to enterprise. Both assumptions share the same fatal flaw - users adopted the products because it was a great deal.

The killer feature was the unsustainable pricing, and we’ll soon find out what happens when you take that away.

The Rise of AI Code Editors

As ChatGPT unleashed a new paradigm of AI interaction, a wave of AI-powered code editors emerged promising to transform how developers write code. These weren't just autocomplete tools; they promised to be integrated development environments (IDEs) and pair programmers that could understand context, explain complex code, debug issues, and write entire functions from natural language descriptions.

Source: Matt Palmer, Replit

The pitch was compelling. It promised “10x developer productivity” through AI assistance handling documentation to debugging. Companies like <u>Cursor</u>, <u>Windsurf</u>, <u>Replit</u>, <u>v0</u>, Convex Chef, <u>Bolt,</u> and Lovable offered seamless integration, context-aware suggestions, and natural language interaction with codebases. The future of development, they claimed, would be a collaborative vibe between human creativity and AI capability. By mid 2025, every company from Airtable to Figma was integrating “vibe coding” into their platform and every post mentioned it.

Within months, the numbers became staggering. Lovable crossed $75M in a 12 months. Windsurf at $100M in 18. Cursor hit $500M ARR in under 3 years. These represented some of the fastest companies to reach $100M ARR in history, outpacing success stories like Wiz and Deel.

Source: Sacra

The playbook seemed perfect. Small, lean teams built AI-powered developer tools that could scale infinitely. Cursor reached $100M ARR with just 20 employees, while Windsurf added $2M in new ARR monthly with a team of 15. This leverage was unprecedented - traditional enterprise software companies needed 5-10x more people for similar growth.

VCs couldn't write checks quickly enough. In two years, Cursor raised $500M at a $10B valuation. Windsurf commanded a $3B valuation before acquisition rumors began swirling. Lovable's rapid rise saw them raise a $200M Series A at a $1.8B valuation 8 months after launching, and Replit reached $3B.

Source: Aventis Advisors

Beneath the surface, a fundamental problem was brewing. The same $20/month "all you can eat" pricing driving this explosive growth masked unsustainable unit economics. Like the neobanks before them, these companies confused price-driven adoption with product-market fit. The assumption was classic SaaS playbook; land users cheaply, then expand through usage limits or enterprise upsell.

This would have worked if they sold high-margin software, but they weren't. They were selling compute, which is more expensive than anyone realized.

Too Many Tokens for Your Thoughts

These companies are selling compute for less than its cost, hoping usage patterns or future efficiencies will help. After an over-engineered deep dive to understand the economics of a code editor (read more in Episode II), the reality became clear; the token unit economics don’t compute.

To understand the unit economics, we focus on the basic unit, the "token." Tokens are the basic units of text that AI models process - they can be chunks of words or parts of words. A token might be a common word like "the," part of a longer word like "comput" in "computing," or a single character. When AI processes text, it breaks it down into tokens, and each requires computational power. The more tokens processed, the higher the computing costs. Even a simple styling change leads to a cascade of AI actions that consume tokens and increase costs.

Replit “effort-based pricing” pricing rationale

A thirty-five token prompt set off a chain reaction inside the AI's brain. First, it read 822 lines of context, then processed it 14 times. It planned, reasoned, generated explanations, and made changes. By the time it finished this supposedly "simple" task, the AI churned through 18,395 tokens of computational power.

At current market rates, this cost the platform $0.62 in raw compute costs. They charged $0.22 for the interaction. That's a 64% loss on every similar request. Even "quick" questions burn through 7,000+ tokens, while debugging sessions can exceed 23,000.

Remember how neobanks discovered they were selling dollars for fifty cents? AI code editors are doing the same thing - faster and at a greater loss. The math reveals an uncomfortable truth: these companies aren't just unprofitable; they're selling compute power at a significant loss, and no clever pricing can fix this fundamental economic problem.

In fact, this problem is worsening due to Jevon's Paradox manifesting in two ways:

First, our expectations of AI capabilities keep expanding. What started as quick responses to coding questions has evolved into multi-minute sessions with thousands of tokens of code changes and detailed explanations. The "magic moment" bar keeps rising.

Source: Metr

Second, while compute costs for older models are decreasing, users demand access to the newest, most capable ones regardless of cost. Ethan Ding's analysis of AI economics notes, "Tasks that returned 1,000 tokens in 2023 now generate 100,000+” - a 100x usage increase driven by expanded expectations and more sophisticated model responses.

Source: Ethan Ding

Remember the viral tweet about AI startup math? He was joking, but the reality reveals something more fundamental. We've built an entire market on the assumption that AI-powered development would follow traditional SaaS economics, which is colliding with the hard reality of compute costs.

The Pricing Model Evolution Crisis

As token costs became clear, companies scrambled to evolve their pricing models. Each iteration revealed the complexity of the unit economics problem and our misunderstanding of AI-powered development.

Phase 1: The $20 All-You-Can-Eat Disaster (2023-early 2024)

Most started with the same playbook: $20/month for unlimited AI assistance. It was a common growth strategy - acquire users now, figure out monetization later. The strategy worked in one sense. Companies like Cursor and Windsurf reached $100M ARR at record pace. But they were effectively paying users to use their product.

Power users exploited this model, consuming 100x the average usage while paying the same flat fee. GitHub’s $20/month Copilot was losing Microsoft $20 per user/month, before reasoning, increased context windows, and vibe code editing came into play.

Phase 2: The Query-Limited Scramble (mid-2024 - mid-2025)

By late 2024, the economics became impossible to ignore. Monthly compute costs per user were skyrocketing and companies faced a difficult choice: raise prices or limit usage, either of which could cause user exodus. Most chose the latter, leading to an era of complex usage restrictions.

Cursor moved to a complex token-based credit system, where different actions cost different amounts. But users revolted en masse at the new pricing that led to escalating costs, leading to a public apology from the CEO.

Many users said they ran out of requests in Cursor rather quickly under the new plan, in some cases <u>after just a few prompts when using Anthropic’s new Claude models</u>, which are particularly popular for coding. Other users claimed they were <u>unexpectedly charged additional costs</u>, not fully understanding they’d be charged extra if they ran over the $20 usage limit and had not set a spend limit. In the new plan, only Cursor’s “auto mode,” which routes to AI models based on capacity, offers unlimited usage for Pro users.

  TechCrunch

Windsurf tried limiting queries, offering "$20 for X queries/month." It was the desperate math of a company realizing its fundamental assumptions were wrong. For what it's worth, Windsurf prices API calls at 20% above the API price at $0.04 (which also means max 17% margins). Their plan pricing of 2,000 prompts for $15 equates to $0.0075 per prompt, costing 0.25 credits, which means $0.03 per credit. So the plan pricing is cheaper than the API pricing? They don't seem to be very good at math.

Source: Windsurf

Phase 3: Effort-Based Pricing Doesn’t Capture Full Effort (mid 2025 to present)

Replit took a fundamentally different approach while others scrambled with query limits and credit systems. They understood that the only sustainable path was to align pricing with computational costs. The result was effort-based pricing - a sophisticated attempt to match user charges with the actual work performed.

Source: Replit

Replit's effort-based pricing aligns costs with usage. They charge based on actual computational work: time spent, actions taken, lines read and changed. It's transparent, fair, and a genuine attempt to match pricing to compute costs.

But here’s the problem: we used Replit’s effort-based pricing model for our cost analysis. Even with the best pricing approach, changing a box shadow color costs $0.62 in compute but generates only $0.22 in revenue. The advanced pricing model still results in ~64% losses on nearly every interaction. Something doesn’t compute. (Note: I used Claude’s pricing for the analysis, the most expensive but the de facto for code editors, so Replit might use cheaper models to make the economics work).

Unfortunately, the company's margins, which should tell the clearest story, are clouded by creative accounting. Companies report 23% to 60% gross margins, but these numbers rely heavily on asterisks - like classifying AI model costs for free users as "R&D" instead of COGS.

Source: The Information

Strip away the accounting magic, and you're left with the reality: these are low-margin businesses pretending to be high-margin software companies. They're selling compute below cost, hoping the economics will improve before reality catches up. Unless they are like Replit and a few others, selling a platform play with higher-margin business lines like hosting and domain purchases to subsidize compute, the math doesn't math.

In Episode II, we’ll do an over-engineered deep dive into the token economics behind AI code editing. We’ll reveal how a simple request to change a box shadow color balloons into thousands of tokens and dollars of compute and why the economics of compute don’t compute.

Full disclosure: I use Kubera and love it enough to write this deep-dive. I get $100 in subscription credits through their referral program (and so do you!) if you sign up through links in this post and reply to their welcome email with my referral, but that's not why I'm writing this. I genuinely believe Kubera is the best solution for complex net worth tracking, and most people haven't heard of it. Consider this my public service announcement.

Early in my career, like any good fintech nerd needing to manage cashflow, I turned to personal finance management tools (PFMs) for the solution. What I didn’t realize is that my journey through PFM tools was the path we all take as we ascend the Over-Engineered Hierarchy of PFM Needs:

• Level 1: The Budget Trackers

• Level 2: The Money Managers

• Level 3: The Wealth Platforms

• Level 4: The Net Worth Dashboard

Level 1: The Budget Trackers

At this stage, I cared most about wrangling my spend into understandable reporting and finding opportunities to save money to enjoy my weekends and time away.

That’s where the Budget Trackers come in. I tried them all, from Mint as the OG budget tracking tool, to YNAB’s overly spreadsheet-driven approach, to Truebill (now Rocket Money)’s subscription management solution. Since my time at Level 1, I also tried new entrants like the beautiful Copilot and the re-imagined Mint as Simplifi. And there are so many more...

Copilot Money Panels

They primarily focus on your income and day-to-day expenses from your main bank account and credit cards. They provide a clear picture of your monthly cashflow so you can see if you’re spending more or less than you can afford.

Ultimately, I discovered too late that budget trackers should only be used for two major things:

1. Finding subscription sprawl: I found Rocket Money most effective at combating subscription sprawl. It has great auto-detection of existing subscriptions, a subscription calendar, and the ability to have the tool cancel subscriptions for you.

2. Quarterly spend reviews: Rocket Money and Copilot excel in categorizing spend to identify causes of spend creep and adjust lifestyle.

Rocket Money’s Subscription Tools

That’s it.

Diligently tracking my daily expenses and categorizing every item that passed my labyrinth of rules didn’t make sense. It wasn’t worth my time to sync accounts when they were out of sync (which happened troublingly often), re-categorize transactions, and update rules and filters.

The brutal truth is that after $100K in annual salary, budget tracking doesn’t make sense. Assuming your total take-home gets shaved by 25% for taxes, that’s $75K. Factor in your rent ($3K) and utilities ($150), and you’re down to ~$35K. With non-discretionary essentials like groceries, transit, and basic essentials, you’re left with ~60% of that or about $20K of discretionary spend. But only the portion of spend on negotiable expenses would benefit from budget tracking, unless you’re looking to reduce your lifestyle enjoyment, which means at most ~50% in the dining/bars, subscriptions, shopping, fitness bucket is negotiable/optimizable. Of that $10K, a budget tracker might get you 5-10% savings, one time (it’s hard to keep finding savings unless you have major lifestyle creep). So after all that work, you might have saved $500 - $1000.

Or, you could set aside 10% of your paycheck to a pre-tax 401(k) which grows at 10% per year in any diversified investment strategy. Without optimizing anything else, you would generate $1K+ per year in investment gains. This is much simpler, requires less mental overhead, and you still end up better off.

After a few years in tech, it’s time to graduate from Level 1 to Level 2 - the Money Managers.

Level 2: The Money Managers

Money Managers expand beyond your checking account and credit cards. They pull in your public investment accounts, joint accounts, auto loans, personal loans, and provide a complete financial picture. They offer planning tools, basic tax optimization insights, and help you think strategically about your finances.

The breadth is valuable when your financial life expands. Instead of asking "am I spending too much on coffee?", you're looking at:

• How are my investments performing across accounts?

• What's my total debt situation, including non-credit card loans?

• How do joint accounts factor into our combined finances?

• Are my investments tax-optimized across accounts?

The names you’ve heard are Monarch Money (useful but expensive) and Personal Capital (great until acquisition). These tools do everything in Level 1 and tackle other parts of your financial life. They also provide a clearer understanding of your financial trajectory and aspects of your overall wealth.

Monarch Money Cash Flow Tracker

But there are clear drawbacks. Investment tracking works for public markets through traditional brokerages, but it doesn’t always cover crypto and alternative investments. Key assets like your first house or car often need manual tracking. Planning is a good start, but if you’re trying to go beyond saving for specific goals to thinking more holistically about your financial goals, these tools don’t have all the answers. And while tax optimization is useful, you’ll still need to file on your own. They also have the same issues with syncing, maintenance, and comprehensiveness as some L1 offerings.

As you continue to achieve, your next level up is Level 3 - The Wealth Managers.

Level 3: The Wealth Platforms

At first glance, Wealth Platforms seem like the final evolution. Companies like Origin, Fifr, Range, SoFi, Wealthfront, and Compound aim to be your all-in-one solution. They do everything Level 1 and 2 tools do, plus advanced planning and analytics, dedicated investing portals, tax filing, and platforms that combine tech, AI, and human advisory.

Each carves out their niche: Origin excels at trust and will services, SoFi brings lending expertise, Wealthfront offers sophisticated robo-advisory, Range provides direct indexing, Fifr automates money movements, and Compound opens access to alternative investments. This feels like the top of the pyramid. Everything in one place, comprehensive service, what else could you want?

But here's the thing: once you reach a certain level of wealth, the all-in-one approach becomes a limitation, not a feature. What you need is simpler but harder to find: a robust net worth tracker that connects to everything, handles any type of asset or debt, and gives insights on your direction. A tool that doesn't try to do everything, but excels at showing you everything.

That's when you graduate beyond the pyramid to Level 4: The Net Worth Dashboard.

Level 4: The Net Worth Dashboard

The final evolution isn't about finding an all-in-one tool. It's about accepting the best-of-breed solution for every aspect of your financial life. I use Robinhood for brokerage, retirement & credit cards, Mercury for banking, Rivet for tax advisory & estate planning. This is the real pinnacle of the hierarchy - not trying to do everything suboptimally, but excelling at the one thing you need: seeing your complete financial picture to make better decisions about specialized tools for each aspect of your wealth.

At this level, you need a powerful dashboard that can track everything without attempting to do everything. That's where Kubera comes in.

Kubera Net Worth Dashboard

First, it has the most complete asset tracking I've seen. Beyond just pulling in traditional investment accounts, Kubera connects to 150+ crypto exchanges, wallets, and DeFi protocols, integrates with data providers for real estate valuations, includes VIN lookups for vehicles, stock options via Carta, private equity and VC investments, and even tracks things other tools ignore like internet domains. If it has value, Kubera can track it.

Kubera’s connections are infinite

The analytics are equally comprehensive. You can model net worth scenarios through retirement, break down asset allocation across platforms, and track performance by asset type over time. I’ve tried every company’s version, from Compound to eMoney, and nothing matches the granularity and simplicity that Kubera provides with straightforward scenario analysis and adjustable inputs.

Kubera’s scenario toggling are the best in the market

They have an interface to ask AI about your net worth and where to optimize. It has fully exportable data to upload LLM-ready JSON for other chatbots like Gemini and Claude.

The practical advantages stack up. The interface is clean and intuitive, especially for spreadsheet users. You can generate clean reports for mortgage lenders' initial underwriting. It supports multi-user access, including advisors. And it stores critical documents like wills, trusts, and other legal documents.

Kubera’s Recap lets you see which assets changed the most each week / month / quarter

The other recent feature is The Billboard, which ranks stocks, funds, crypto, banks, brokers, and exchanges in real-time based on subscriber usage. There is also a percentile calculator to compare your net worth to others on the platform.

Move aside Billboard Top 40, there’s another Kubera Billboard in town

What sets Kubera apart is what it doesn't do. It doesn't attempt to offer overly general tax filing service, trust services, or try to be your everything platform. At $249/year, it's focused purely on being the best possible dashboard for complex financial situations.

Like Maslow's Hierarchy, the journey through personal finance tools leads to self-actualization. You start with basic needs - tracking spending and finding leaks with budget trackers. You progress through psychological needs - understanding your broader financial picture via money managers. You reach for self-actualization through all-in-one wealth platforms. But true financial self-actualization comes from accepting that you need the best tools for each job, unified by a single clear view of your wealth - the net worth dashboard. The peak is about seeing everything clearly so you can make better choices about what you actually need.

They’re wrong, and this is another attempt at thinly veiled sexism, hidden behind a misunderstanding of business metrics. Let’s dismantle their weak argument piece by piece and learn about the WNBA and the basketball business.

The WNBA makes more than enough money to pay its players, especially considering recent deals.

Let's start with the basics around whether the WNBA has enough money to pay them more. The misunderstanding comes from an excessive focus on prior year performance to judge future year pay.

The WNBA reported $200M in 2023 revenue. With 40% year-over-year growth based on the women's basketball industry, we're looking at ~$400M by end of 2025. That's significant and just the beginning.

Because last month, the WNBA signed a groundbreaking $2.35B media rights deal spanning 11 years - roughly $213M annually starting in 2026. ION, the network airing games, saw 23M unique viewers alongside an estimated 54M in the US. This is particularly impressive given that, as an NBA subsidiary, the WNBA can't even independently negotiate these deals. Yet they secured the largest media rights deal in women's sports history.

But wait, there's more. The league is adding five expansion teams over the next five years. They are: Golden State Valkyries (2025), Toronto Tempo (2026), Portland Fire (2026), Cleveland (2028), Detroit (2029), and Philadelphia (2030). Each team will pay a $250M franchise fee. That's $1.25B in expansion revenue, or $250M annually.

These new teams will generate additional league revenue. The current 13 teams average $30M each in annual revenue ($400M / 13). Adding five more teams at similar performance increases annual revenue by another $150M.

Let's calculate the annual run-rate revenue:

• Current league revenue: $400M

• Media rights: $213M

• Annual expansion fees: $250M

• New team revenue: $150M

Total run-rate revenue: $1,013M

That's over $1B in annual revenue - and growing fast. Suddenly, the $50M annual loss critics cite seems less relevant. The available pool for player salaries could be dramatically higher without threatening the league's financial position.

Don’t believe me? Let’s break down the WNBA P&L today with its 13 teams to get to $56M in losses in 2025 (growing the $40M losses reported in 2024 by the same topline growth rate and assuming no efficiencies):

• Revenue: $400M

• Player Salaries: $40M (using ~10% of revenue as is widely reported)

• Non-player Opex: $400M (the plug to reach $40M in losses)

If we extrapolate this to 18 teams (adding 5 expansion teams), that's about $553M in non-salary operating expenses against our $1,013M run-rate revenue. We could increase player salaries from the current $55M run-rate (for 18 teams) to $250M - a 400% growth rate - and still maintain operating margins similar to the NBA.

We haven't even factored in several major growth opportunities:

• Season extension potential: The WNBA currently plays 44 games per season compared to the NBA's 82. Extending the season could significantly increase media rights and ticket revenue.

• Growing attendance: Teams are already hitting capacity constraints with 11,000+ fans per game (double 2022 levels). This is driving even higher ticket sales and league revenue.

• Social Media and OTT: The WNBA saw 2B views on social media platforms (4x last season), a 366% growth in WNBA’s OTT League Pass subscriptions, and a 2.5x increase in WNBA App MAUs.

• International expansion: The WNBA now has a 21% international roster, distribution deals for the UK, China, Brazil & the Philippines, and a Toronto expansion team.

• Tournament expansion: The league is exploring additional tournament play and extended playoff formats, following the successful NBA In-Season Tournament model.

• Merchandise: Since 2023, WNBA merchandise sales are up 600% and partner store sales are up 2033% year-over-year.

These initiatives could push revenue significantly beyond $1B annually.

Suddenly, the $50M annual loss that critics love to cite seems less relevant. The player salaries pool could be dramatically higher without threatening the league's financial position.

The money is there. The players’ pay isn’t.

The WNBA is a massively successful growing franchise whose value is understated if you consider net income alone, so it can pay players fairly.

Let's discuss the value creation standpoint because investments in player pay and other expenses create enterprise value. On the topic of enterprise value, the WNBA's 13 existing teams are now worth $3.5B combined (2x in the last year):

Add the five expansion teams at $250M each, and we're looking at a $4.75B league. Not bad for a "money-losing" venture.

Speaking of losses, the WNBA lost about $250M ($10M annually) over its first 25 years, plus another $160M in the last 4 years post pandemic with all stars entering the league. So that's a total of $410M of losses to build a $4.75B asset. In venture capital terms, that's better than a 10x return. Any VC would be eager for those numbers.

The "unprofitable" argument gets sillier when you look at the broader market. Over half of Russell 2000 companies operate at a loss. Some of the most valuable ones - Amazon, Uber, Airbnb - were "unprofitable" for years while building massive enterprise value. Tesla didn't turn a profit until 2020, yet created hundreds of billions in shareholder value.

Focusing on net income alone misses the point. The WNBA isn't just growing; it's creating enormous value. A 10x ROI is a no-brainer for the NBA. Paying players fairly to secure and grow that value is an easy decision.

The WNBA should pay its players fairly to compete with other leagues.

Let’s start with the comparison to the NBA, its parent and most comparable organization. Of course, the NBA is significantly larger, and we can adjust accordingly. The NBA generates roughly 10x the revenue of the WNBA, a still growing franchise. But this 10x revenue difference translates to a 100x gap in minimum salary ($1.1M NBA vs $65K WNBA) and a 65x difference in average compensation. How does that math work?

Of course, the math still doesn’t justify the WNBA players’ low pay as a percentage of revenue. The WNBA allocates just 10% of its revenue to player compensation, while the NBA shares over 50%, comparable across leagues. This disparity is unjustifiable.

What’s also troublesome is that only the WNBA can’t justify higher pay when other American women’s leagues can. The new Unrivaled 3on3 Women’s League pays $220K for 8 weeks - nearly matching the WNBA’s maximum annual salary of $235K. Read that again: The average player can make almost as much in 8 weeks in Unrivaled as the best WNBA player.

Even more concerning is the global competition for talent. Of the 11 major international women's basketball leagues, 70% pay higher average salaries than the WNBA. Top players can earn $1M annually in China.

When we can fairly pay players and we're losing talent to international leagues that do, maintaining the current compensation structure isn't just unfair - it's unsustainable. The WNBA can't build a world-class league while paying world-class talent below-market rates. It's falling short of every benchmark right now. And it needs to change.

So the WNBA makes more than enough money to pay its players, especially with recent deals. It’s a massively successful growing franchise whose value is understated if you consider net income alone, so it can pay players fairly. And the WNBA should pay its athletes fairly to compete with others domestically and internationally. The players deserve all of the raises and increases to pay that they get in the upcoming conversations this year. It’s a start.

I recently joined my good friend Nico Simko, Co-founder & CEO of Clair, on HR Tech Trailblazers to discuss my journey in fintech and my thoughts on all things HR Tech.

While I definitely didn't plan it this way, working across early-stage startups, hypergrowth companies, profitable trading firms, and now public companies has given me a unique perspective on how we can expand financial access at scale. A few highlights from the interview:

1. The evolution of payroll platforms is happening now: We're watching payroll companies transform from basic systems of record into sophisticated platforms for financial services innovation. Their thoughtful approach to data sharing, while sometimes challenging for fintechs, creates the foundation for sustainable innovation that protects both employers and employees.

2. Earned wage access isn't just a nice-to-have, it's a necessity: When we talk to users at the fintechs I've worked at about what matters most, it consistently comes back to liquidity. With 40%+ of Americans wanting faster access to their earned wages, this is more than a convenience and is all about financial security. At Block, we're thinking deeply about how to solve this at scale.

3. Ecosystem plays are powerful but complex: Connecting Square Sellers with Cash App users and leveraging our unique ecosystem creates unique opportunities to help both businesses and their employees. While the execution is challenging, I'm convinced that these interconnected solutions will define the next generation of financial services.

Thank you to Nico Simko, Karen Yung and the team that made this happen!

When the announcement of the transaction occurred, politicians, pundits and regulators claimed it would be detrimental for consumers because:

1. It created another giant bank that would eliminate banking competition.

2. It would limit consumer access and services.

3. It would exacerbate the network duopoly problem.

These arguments are not accurate. Now that the transaction is completed and implementation is starting, it’s time to revisit and break them down.

Myth #1: Capital One + Discover create a giant bank that eliminates competition.

Let's address the elephant in the room: the "too big to fail" argument. Critics cite the combined assets of Capital One and Discover as making the 6th largest bank in the U.S., proving regional banks will be crushed by this acquisition.

The argument relies on a flawed interpretation of perfect competition in banking. In this interpretation, the best economic outcomes result from many banks with no single one influencing the market price of banking or access to services.

But this view needs context. We don’t operate in a world of perfect competition for banking services but one dominated by systemically large banking institutions that can’t fail. Sure, Capital One becomes one of the largest banks compared to regional institutions, but what about the “little old” JPMorgan Chase, Citi, and Wells Fargo that CO-DF will also be fighting? Even combined, CO-DF will have 70% fewer deposits than the smallest of the Big 4 systematically important financial institutions.

The "too big" narrative ignores a key reality of modern banking: scale is necessary to deliver competitive services. Without sufficient scale, banks can't afford to provide scaled banking services, leading to worse service, limited underwriting, and higher customer costs. Larger banks have ~35% better loan economics, 30% lower expenses, and are better regulated than smaller ones.

Overbanked

posts.interspace.ventures/p/overbanked

We're not creating a new megabank. Instead, we're enabling an existing large bank to better compete with larger incumbents. That's how you drive real innovation in financial services. Sometimes you need a Goliath to fight other Goliaths.

Myth #2: The Capital One + Discover merger limits consumer access and services

Critics of the merger are so focused on the "big bank" narrative that they are missing how the customer bases and product offerings complement each other. This merger is about addressing gaps in each company's portfolio.

Capital One has mastered premium credit cards and tech-forward banking but struggled to expand their debit and everyday banking presence. Meanwhile, Discover has built a loyal following with cash-back rewards and US-based customer service but never cracked the premium card market.

This combination enables a unique full-service financial platform that can compete across all segments while maintaining the customer service that made both brands successful. Capital One's efficient self-serve technology and Discover's renowned US-based customer support create a service model that JPMorgan Chase and Wells Fargo cannot match.

The technology angle is crucial. Over the last decade, Capital One spent on digital transformation to disrupt itself before fintech could. It pushed for cloud migrations, labor force transformation, and a focus on using data to self-serve, personalize, and better serve its customers.

Sandeep Sood, former VP of Engineering at Capital One, says it now has the best technical infrastructure of any US financial institution of similar size - it’s “the world’s largest fintech.” Capital One is bringing that expertise to overhaul Discover’s legacy technology and infrastructure, anticipating ~$3B of annual operating and network synergies.

Critics arguing this hurts competition have it backwards. The combined entity will force other banks to improve their digital capabilities and customer service to keep up. That's how real competition works; it drives innovation and better customer experiences.

Myth #3: Capital One + Discover would worsen the network duopoly.

Let's start with the facts: Visa and Mastercard dominate payment networks with a combined market share of over 94% of US credit cards. The remaining players - Amex, Discover, and regional networks - fight for the leftovers. The network effects make it nearly impossible for smaller players to compete.

Think about it this way: A new payment network needs both merchants to accept it and consumers to use it. Merchants won't accept a network without consumer volume, and they won't choose a card that's not widely accepted. It's the classic cold start problem that has protected Visa and Mastercard's duopoly for decades.

This is why we so desperately need competition and a large player to compete. Small networks have failed to make meaningful inroads. A recent Fed study shows that despite 15+ minor networks in the US, they've had minimal impact on interchange rates or merchant acceptance (I'll bet only a few of you fintech nerds know Jeanie, Shazam, Maestro, NYCE, START, Culliance, Accel or the one I just made up because the brands are so unfamiliar). The painful truth is that you can't fight network effects with subscale solutions.

Capital One + Discover change this dynamic. Even combined, they will still be smaller than Visa or Mastercard, but they have the scale and resources to compete. Here's why this matters:

1. Capital One has massive consumer reach through its credit card base.

2. Discover brings an established merchant network and payment rails.

3. Together, they have the technology and capital to increase acceptance.

Capital One + Discover could be the first real challenge to the payment network duopoly in decades, benefiting everyone (including even Visa & Mastercard who thrive in competition but may have forgotten along the way).

The Real Opportunity: Scalable Banking-as-a-Service

While everyone's focused on traditional banking, they're missing the biggest opportunity. This merger positions Capital One to lead the Banking-as-a-Service (BaaS) space. Here's why that matters.

The BaaS market is fragmented between smaller players like Green Dot and regional banks monetizing their charters. But none have the technical capabilities and scale to serve major fintech partners effectively. Capital One + Discover changes this.

The banks that power your favorite fintechs

posts.interspace.ventures/p/the-banks-that-power-your-favorite

Consider the assets this combination brings to BaaS:

• Capital One's automated underwriting capabilities

• Discover's payment network and settlement infrastructure

• Combined fraud detection and risk management systems

• Massive data sets for better decisioning

• Technology-focused approach to banking services

This lays the foundation for next-gen fintech partnerships. Imagine:

• Co-branded cards with automated instant decisioning

• Built-style partnerships leveraging both debit and credit rails

• Embedded banking solutions with real-time fraud detection

• Direct access to a competitive payment network

We've already seen hints of this strategy. Capital One has been investing heavily in API infrastructure and cloud capabilities. Discover has been quietly building its digital banking platform. Together, they could offer what every major fintech wants: a full-stack banking partner with modern technology and competitive economics.

This isn't speculation. The investor presentation highlights technology investment and partnership opportunities. They're thinking beyond traditional banking - they're building the infrastructure for the next generation of financial services.

The Big Picture

The Capital One-Discover merger isn't just about creating another big bank. It's about building the first modern financial infrastructure company at scale. This combination has the potential to:

1. Force traditional banks to compete and introduce competition among the Big 4 SIFIs.

2. Increase customer access and services while supercharging tech.

3. Break the payment network duopoly for good.

My fintuition tells me we'll see this deal as a turning point in financial services, not because it created another big bank, but because it created the first real challenger to decades-old industry structures.

The critics aren't wrong that this deal will create a larger bank. They're wrong about why that matters. Will Capital One + Discover reshape financial services? That depends on execution. But for the first time in decades, we have the potential for real change in payment networks, banking infrastructure, and financial technology.

Sometimes you need to build something big to break up something bigger. This deal might just do exactly that.

Interspace VC Index - Q2 2025 Additions

If there’s a household name in global fintech backing, it’s DST Global. Yuri Milner’s brainchild, spun out of the Mail.ru empire in 2009, oversees over $50B in assets. Their $4B 2021 fund has allowed them to discover and double down on winners like Chime, Robinhood, Nubank, and Stripe. Few firms wield that level of firepower or conviction.

Next up, Flourish Ventures operates on a thesis-driven model from eBay founder Pierre Omidyar. Their small global teams focus on Series A and occasional B, pursuing high-impact opportunities that blend mission and scale. Recent highlights include leading a Series B for Reserv and enabling BaaS with Unit.

Formation just launched their $25M inaugural fund and they’re already busy. Strada alum Solly Garber and Floodgate alum Leeor Mushin are writing the first checks ($600K–$900K) to pre-seed founders with big visions including companies like Project 11. Their playbook: cap the portfolio around 30 companies, lean in early, and stay aligned with teams building fintech’s next infrastructure layers.

Of course, Foundation Capital hardly needs an introduction in Silicon Valley. Now on Fund 11 ($600M) and managing $4B across investment partners, they split their thesis between financial services/crypto and enterprise AI/SaaS. Partner Nico Stainfeld, Zach Noorani & Gracie Zaro have been developing their new Fintech AI Build accelerator which is accepting fintech applications for equity-free, stocked with cloud credits from OpenAI to Anthropic, coast-to-coast workspace.

Vine Ventures brings a New York-Israel flavor to early-stage investing. Since closing a $140M second fund, the small team led by Eric Reiner, Dan Povitsky & Demren Sinik has backed founders with $1-$10M checks and continued to elevate companies beyond capital. Their portfolio includes marquee fintech: Carta, Ramp, MANTL and counting.

Finally, White Star Capital deepens our venture capital coverage. Since 2014, they’ve quietly grown to $1.4B in AUM and led Series A/B rounds for fintech and B2B SaaS including early investments into Percent, OatFi, and BoomFi. Their New York team combines ex-JPM tech bankers, quantitative analysts, and product-obsessed associates who scout the next wave of fintechs.

That’s a wrap for this quarter’s Interspace Index updates and here’s the full list - Interspace Index - Fintech VCs | Updated July 2025. Think we’ve missed a fintech VC? Reply here or DM me @heysamir_ on Twitter.

I’ll refresh the Interspace Index quarterly, sticking to our criteria: a primary focus on fintech, at least one fintech investment in the past year, or genuine community-builder credentials via founder nomination. Onward!

Yeah, this was awkward. The same day I posted about Wells Fargo's deal, Bilt announced a $250M round valuing the firm at $11B with $100M of strategic capital from United Wholesale Mortgage and WF is ending the deal early. Was I off the mark?

Maybe it wasn’t actually about economics.

The early termination of the Wells Fargo deal initially confused me. I still firmly believe that economically this wasn’t a bad deal for Wells Fargo, especially since the core criticisms came from a WSJ article that misinterpreted Wells Fargo executives’ statements. One example of many I detailed in my post:

WSJ said Wells Fargo assumed 50-75% of balances would be revolving. However, for the credit score Bilt targeted, the revolving balance is only 20-30% or less. Given the size of an average rent payment in someone’s monthly spend, getting to 50-75% would mean targeting subprime creditors (which aren’t eligible for the card at <580) OR assuming the core demo can revolve on rent payments (which isn’t possible or this product wouldn’t exist).

Instead, Wells Fargo likely assumed 50-75% of non-rent payments would revolve, amounting to ~10-15% of total balances. This assumption aligned with the actual metrics from the partnership, indicating the deal wasn’t unexpected economically. There’s unfortunately a few other misinterpretations in the article like this which when reinterpreted make it clear that the partnership economics were actually sound.

But on a basic level, Wells Fargo also acquired ~675K cardholders for ~$120M/year. If you're a bank, you'd pay $180 CAC any day since customer acquisition is often $1K+ given the average bank relationship lifetime is 10+ years. Remember Chase CEO Jamie Dimon's quote:

One of the fictions here is that the marketing cost ... gets booked over 12 months. The benefit of the card gets booked over 7 years. The card was so successful it cost us $200 million, but we expect that to have a good return on it. I wish it was a $400 million loss.

Of course, the business model was potentially different than what Wells Fargo imagined when it underwrote the deal. Even this is a stretch because the numbers show card spend was never more than ~25% of overall rent payments and rapidly went down so it couldn’t have been a surprise. In any case, with $1.4B in outstanding balances per WSJ, and ~675K members (15% of 4.5M homes on the platform), the $2K average revolve might not have looked appealing from a traditional credit card economics standpoint, but it was about much more.

Instead of a card with high revolving balances and a traditional credit card economics model, they were getting a customer acquisition engine and a lower-touch, lower brand engagement sponsor bank model. While Wells Fargo could have run a sponsorship model that many banks have minted billions with, it wasn’t ready to take on this model that Cardless, Bilt’s new card partner and behind Coinbase’s new Bitcoin Amex card, will now monetize. The economics are viable, but some banks must be ready to take a less prominent role if they can't bring the rewards network.

The banks that power your favorite fintechs

posts.interspace.ventures/p/the-banks-that-power-your-favorite

But Wells Fargo wanted to increase their brand equity. They wanted what Chase got with Chase Sapphire Reserve - the Wells Fargo Bilt Card. Instead, it became the Bilt Rewards Card with a tiny Wells Fargo reference. That's Wells Fargo's loss and an opportunity for First Electronic Bank, the sponsor bank powering Cardless.

Don’t be surprised by more new entrants in this space. When valuations reach decacorn status, startups will emerge. I see potential entry points from one or all of the following:

1. Chase entering directly and leveraging its in-house and partnerships-driven rewards network or issuing via a co-brand.

2. Amex (with its substantial Offers rewards platform) entering via co-brand partnership to launch this partnership, potentially issuing through Cardless or Imprint.

3. A startup partners with either to issue a “points on rent + points on mortgage” card. Haven, Mesa and Further are already in the mix. It’s going to grow.

It’s a great time to enter the space Bilt has paved, given the expected expansion into condos & HOAs, student housing, and mortgages. It’s a different model, but one that can work well; consider the co-brand cards from Chase or Capital One.

Share this post from Interspace

Maybe it was a technology and CX problem.

Of course, to be the bank behind the fintech, you need to be built to engage with a tech-forward company and a tech-savvy userbase. It’s unclear if Wells Fargo was ready for this experience. Its underlying technology and integrations were a sore spot for Bilt with complaints about customer support and card repayments leading to a low 1.7/5.0 rating on Trustpilot:

Bilt would have wanted to disintermediate Wells to improve the customer experience, but this would have made it harder to identify Wells Fargo behind the app. Given Wells Fargo’s regulation, it likely made sense to reduce and retrench the business rather than overhaul the legacy tech infrastructure.

Maybe it was fear of competition.

While Wells Fargo let people believe the end of the deal was an economics story, it was probably bristling at Bilt signing up another mortgage provider, United Wholesale Mortgage, which invested $100M of strategic capital into the latest round.

United Wholesale Mortgage (UWM) is a wholesale mortgage lender that underwrites loans for independent real estate brokers. Since 1986, UWM has been the largest mortgage lender in the U.S. Brokers originate loans for home buying, and UWM underwrites and funds them, then sells them to secondary markets of mortgage servicers. They also provide tech, training, and support for mortgage brokers.

Bilt was always going to build a marketplace because multiple providers make a network compelling. But if WF had to compete on price, they knew they wouldn't always win. Perhaps it was better to exit the partnership rather than becoming commoditized in a race to the bottom on pricing.

The breakup was better for everyone involved. Bilt gets a tech-forward partner that doesn’t mind taking a backseat on brand because it monetizes on transaction volume and deposits, not customer acquisition and cross-sell. Bilt adds UWM to supercharge its mortgage and home-related services expansion while Wells Fargo rethinks its strategy.

If you Bilt it, they will come (with high-flying valuations)

Finally, a word on Bilt's valuation. I assigned a lofty 15x multiple on $200M of revenue. Should the valuation be $15B when the company reaches $1B in NTM revenue? No, not when it is operating what will be primarily a rewards network at mature scale with more measured growth prospects.

An excellent article from my friend Michael Bloch, Partner at Quiet Capital, states that Delta made ~10% of its revenue (almost $6B in absolute and up to 50% of its profits) from co-brand partnership with Amex, a useful proxy for Bilt's rewards network business model.

Let's assign a generous $20B to the Delta rewards network with Delta's market cap of ~$40B. That gives rewards a 4x multiple. So what does that mean for Bilt, who is expecting $1B in revenue by Q1 2026? That takes us back to ~$4B for Bilt. Back in my day, $4B was impressive though, but maybe not good enough for Bilt's lofty aspirations.

Let’s see how Bilt’s valuation develops as it matures into a potential IPO-ready company in a few years. Perhaps Bilt’s high-flying valuation will be a rising tide to lift all boats, or maybe we’ll see the true valuation rationalize with time.

Bilt is back in the news, but not for a misunderstood bank partnership that everyone wrote off as unsustainable. It’s for rumors they’re attempting to raise at a whopping $10B valuation, making them unofficially a decacorn if true. While I agree the valuation is generous, I disagree with the critics that Bilt struck a bad deal. By extension, it’s clear the Bilt partnership with Wells Fargo and Mastercard is a smart and replicable one, and startups should copy it (WSJ gave you the playbook, and Bilt’s recent update to valuation shows it works).

They Bilt a Housing & Neighborhood loyalty network

To understand Bilt, we need to explain its platform and business model. At its core, Bilt is a rewards network and loyalty program. It is an enhanced and streamlined affiliate network.

Here’s how it functions:

1. Merchants go to Bilt for access to their userbase.

2. Bilt works with the merchant to structure an appealing rewards offer, like significant cashback, to bring users into the store.

3. The merchant agrees to provide a 10% bounty. Bilt offers users 8% back in points, keeping the 2% on every transaction.

Everyone wins. Merchants attract more users at a comfortable price. Users get significant cashback and discover new merchants. Bilt makes a tidy sum at nearly 100% gross margin.

The economics can be better for Bilt and the merchant because, while the headline is perceived as 10% cashback, it’s actually 10 points per dollar and the business of points is a whole different ball game. They must be redeemed to matter and there’s a drop-off of up to 40% in points spending. The points to cashback transfer is only 1:0.5 so 1 point is worth $0.50 cashback. A merchant can offer 10% cashback and be comfortable knowing it’s only 3-5% cash out the door.

The Business of Free

posts.interspace.ventures/p/the-business-of-free

They Bilt a viral user acquisition engine

How did they build a strong userbase to get merchants excited about marketing to them? That’s where Bilt’s value proposition came in - points on rent, for any building no matter what. Recently, the CEO, Ankur Jain, explained this:

At its core, we started Bilt to reward people on their biggest monthly expense, which is rent. And it’s funny because it’s one of those things where people have totally taken it for granted. But you spend 30%, 40%, in some cities 50% of your income on housing costs every single month, and, yet it’s the one category that hadn’t really been modernized to the way that the rest of the payments ecosystem has evolved. I mean, if you’re under 35, the only reason you probably have a checkbook still is for rent. So can you digitize that, and can you do what other payment worlds have done, and bring rewards into the experience?

First, let’s set the economics aside for a moment to understand mechanically how rental payments via Bilt work. There are two ways Bilt makes that happen:

1. Direct property integration: Bilt integrates directly with the rental property. This enables seamless, instant, on-time payments for landlords while providing a feature renters want. Bilt monetizes payment processing like a standard processor.

2. Indirect property integration: Bilt offers a smart workaround for properties that haven’t integrated with Bilt. They set up a shadow bank account for users where the company provides just-in-time funding for rent payments off their own balance sheet and debits your credit card. While it doesn’t generate economics on its own, it expands their reach and increases user acquisition while avoiding the credit card processing fees landlords charge.

Together, Bilt can credibly market a claim of providing points on rent for any building that accepts bank payments.

Share this post from Interspace

They Bilt valuable (& replicable) strategic partnerships

Points on rent is an amazing benefit, but so is 100% cashback. It doesn’t matter unless you can build a sustainable business model around it. That’s where Bilt’s two key strategic partnerships come in.

Mastercard

Mastercard has stayed out of the headlines, but it is a key part of the strategic partnership ecosystem Bilt created. Normally, rent payments over the network incur substantial network fees that make them unsustainable (hence the high credit card processing fees or inability to use your credit card to pay a landlord).

But Bilt struck a favorable deal with Mastercard to reduce or waive typical network fees. This was likely in exchange for a treasure trove of data about renters’ financial behavior that networks normally don’t see with ACH/wire transactions. Payment networks talk about “new flows” to stay ahead of any payment flows that might reduce dependence on them, like real-time payments (RTP), FedNow instant payments, A2A transfers, or BNPL. You can bet that Mastercard values this data and would pay for it.

Wells Fargo

Restating my 2024 over-engineered year in review, the Wells Fargo partnership is the most well-known and controversial from the Bilt saga. The drama started in June 2024, the WSJ reported that Wells Fargo was losing millions monthly on its partnership with Bilt Rewards, and it turned our ecosystem upside down. The gist of the argument was the following (as summarized by Sheel Mohnot on X):

1. Wells Fargo assumed the Bilt card would be the top one for renters with balances, but most users use it only for rent and five required transactions.

2. Wells Fargo assumed 65% of card purchase volumes would be non-rent, generating interchange revenue. However, the reality was over 65% rent-related, earning very little (<40bps gross and effectively negative with cashback).

3. Wells Fargo assumed 50-75% of balances would be revolving and earn interest, but most don’t carry a balance and only 15-25% is revolving.

4. Wells Fargo is paying ~0.80% to Bilt on rent payments and hoped to cross-sell its mortgages, but that hasn’t happened yet.

The early WSJ takes that the cash back on rent payments model doesn’t work are wrong. Both WF and Bilt denied issues in the contract, which extends until 2029. It is also funny that a WSJ reporter talked to only Wells Fargo about the contract negotiation and took their word on opinions that could just be negotiation talking points. Didn’t they wonder why WF was eager to “leak” this?

Here are a few counterpoints to the article’s main points:

1. We’re 2 years into a 7-year partnership. It’s too early to measure top-of-wallet success. Any increase in spending is a success, and they have the actual rent payment on the card (if rent isn’t your biggest monthly spend, I want to be you).

2. Wells Fargo could not have thought this fact was true and the statement was misinterpreted. They said 65% of non-rent card purchase volumes wouldn’t relate to rent-related payments (e.g. utilities, maintenance, internet) and it was misstated. Wells Fargo knew the majority of card spend would be rent-related as that’s the case for every individual anyway (~50%+ spent on rent/housing). People were eager to criticize Bilt and assume Wells Fargo was foolish, but remember “Banks are slow, not stupid.”

3. I can’t believe this was true and this is another misinterpretation. The actual statement was 50-75% of non-rent payments would revolve. For the credit score that Bilt targeted, the revolving balance is only 20-30% or less. Getting to 50-75% would mean they’re targeting subprime creditors which aren’t eligible for the card (<580) OR they assume the core demo can revolve on rent payments (which isn’t possible or this product wouldn’t exist).

4. We’re two years into a 7-year relationship. At the time of the article, Bilt didn’t have an upsell motion for homebuyers. Within months, they launched a program to earn Bilt points on home purchases and revamped tools to help renters estimate monthly payments including taxes and insurance. This is just the beginning, and any smart fintech nerd can see many ways to expand the upsell opportunities.

The article hinged on a supposed $10M/month loss, but for a bank making $20B/quarter, this is a rounding error. Bilt may not succeed, but judging them by this article is the wrong way to evaluate them.

Remember Jamie Dimon’s quote on the Chase Sapphire Reserve Card’s success:

One of the fictions here is that the marketing cost ... gets booked over 12 months. The benefit of the card gets booked over 7 years. The card was so successful it cost us $200 million, but we expect that to have a good return on it. I wish it was a $400 million loss.

The article increased interest in consumer credit in the mortgage space. We saw mortgage servicing platform Haven launch in September 2024, Mesa launch in November 2024, and Further to help aspiring homeowners. This model works, it’s replicable, and startups are already doing it.

They built a self-sustaining virtuous cycle.

With these three pieces, they have what every startup wants - sustainable product market fit with a self-reinforcing virtuous cycle:

1. More users join Bilt for unique rental/mortgage benefits and a network of merchant rewards.

2. More merchants join the platform and offer more rewards due to the growing user base.

3. Bilt can negotiate better strategic and financial partnerships to strengthen the business model and unit economics.

When this works, it benefits all stakeholders and mints a sustainable, competitive business model. The market will reward you handsomely (perhaps too handsomely) for building in this space, as seen by Bilt’s jaw-dropping 50x revenue multiple implied by their latest rumored round, and you can do this profitably as suggested by the rumors of Bilt having been profitable since their prior round.

All that said, this valuation is still overpriced. Bilt is a loyalty rewards network, and we have many. It makes no money on the credit card (as was always their plan), so why are we valuing it like the next big neobank? They have an amazing business that should get a ~15x revenue multiple at most, taking them to $3B (which was their last round). If someone offers 2% back on rent + 1% back on mortgage and uses a company like Percents for merchant-funded rewards, I don't see how Bilt competes except on rewards pricing in a downward trend.

I credit the Bilt team for building (bilting?) a network effect and competitive moat that is hard to dethrone. But let’s remember that this network effect makes Bilt’s business model hard to replicate, not special partnerships. Startups in this space have a shot at expanding this concept and building on or adjacent to Bilt. There’s room for many winners because your competitor isn’t Bilt, but the 99% of people not optimizing their rent and mortgage spend. I’m excited to see what more fintech nerds build and how users continue to benefit from this innovation.

Share this post from Interspace

Subscribe

Last month, during the Chime IPO week, I was in Atlanta for a team offsite. On the spur of the moment, I decided to fly back for IPO day and caught an early morning flight that arrived at JFK at 8:30 AM ET. With ~1hr before opening bell, I needed to get to Times Square as quickly as possible. A train would take at least another hour and a half and would still require another transfer. A car might have made the time, but it was rush hour and getting into the city would be difficult.

So I decided to take a BLADE helicopter from JFK to their helipad on the west side of Manhattan. For those unaware, BLADE is modern helicopter commuting:

BLADE is a technology-powered, global urban air mobility platform committed to reducing travel friction by enabling cost-effective air transportation alternatives to some of the most congested ground routes in the U.S. and abroad. Our long-term goal is to make aviation more accessible by preparing for the adoption of EVA (Electric Vertical Aircraft) quiet, carbon-neutral and cost-effective aircraft that are currently being developed by our investors and partners.

The helicopter service is straightforward. It has a maximum of 4 passengers per flight, Uber-like booking, and a modern helicopter fleet. Trips take ~15 minutes from hangar to touchdown with a 5-minute on-flight journey that also serves as a helicopter tour of your favorite cities. They will pick you up from the airport, drive you to the helipads, and offer dedicated private cars from the well-located helipads to the city (for an additional fare).

My flight took 5 minutes. Then, another 15 minutes for the Uber into Times Square. I arrived just in time to see the company I had been with for 1/3 of my career finally list.

Share this post from Interspace

"It's Too Expensive"

This is the most common pushback, but it misses the forest AND the trees. For me, there was no other way to make the time given my flight, so the marginal ~$100 for a priceless event is obvious. Practically, to make the time any other way, I would need an earlier flight the previous day which would cost over ~$150 extra. So it was the cheapest option for me, empirically.

But I can prove it’s the best value option even on non-IPO days. Here’s the math:

(1) Time Saved

The average BLADE trip takes 15 minutes door-to-door, while a cab ride from JFK to the same location takes about 2 hours.

Using an average salary of ~$200K for tech workers with 10 years of experience (likely low but fine for this purpose), the marginal minute value is ~$1.80. Saving at least ~100 minutes of travel time, the value is conservatively ~$180.

The $195 BLADE fee includes tip, so the time saved covers the BLADE’s $163 cost. But we’re not finished.

(2) The Helicopter Tour

When booking a BLADE, you don’t consider the benefit of the breathtaking helicopter tour of the city, but you immediately experience it. I’ve flown into and out of the city, and the experience is equally impressive. Normally, people pay for that privilege.

I estimate the BLADE flight time to be ~5-7 minutes. At 5 minutes, it should be worth at least $260 / 15 × 5 = $86. They aren’t highlighting things for you, but that’s what your phone is for!

Economics

We pay $195 per seat for $266 value via time saved ($180) and extra benefits like a helicopter tour ($86). That seems like a smart financial decision. Suddenly, that helicopter doesn't look so expensive. But is it safe?

"It's Not Safe"

This is usually the strongest pushback against helicopters. However, despite the perception, helicopters are much safer than they appear.

Safer than Air Travel

First, let’s compare flying on a commercial helicopter like BLADE to other forms of air travel. BLADE helicopters are considered civil helicopter operations, including sightseeing tours, charters, air ambulance, news agencies, film agencies, and agricultural operations. In that group, the fatality rate is ~0.62 per 100,000 flight hours.

For commercial non-sightseeing operations like BLADE, safety is even more evident. According to the US Helicopter Safety Team, there hasn’t been a helicopter accident in New York City for as long as the FAA has recorded the data (at least 18 years). So the rate is specifically 0 for commercial non-sightseeing operations. Compare that to other forms of air travel:

Safer than Trains

An alternative to flying a helicopter is to use a train, which should be extremely safe given the infrequency of major accidents, right?

Not exactly. Train and other wheeled travel accident rates are conveyed in accidents per million miles driven rather than per hour traveled, so we need some math. Train accident rates have been rising and average ~1.4 accidents per 1M miles traveled. Trains travel ~60MPH inclusive of stops (which is why any train on the eastern corridor is no faster than driving), so 1M miles translates to ~17,000 hours of travel. That means trains are closer to 8.3 fatalities per 100,000 hours. Yikes.

Safer than Cars

Lastly, let’s compare cars like the Uber or taxi I’d normally take. Cars aren’t safe. But just how unsafe surprised me when we compare them using the same metrics as helicopter safety.

At ~40MPH for cars, 1M miles translates to ~25,000 hours. So, 13 deaths per 1M miles driven is 52 fatalities per 100,000 hours driven.

Will autonomous driving save us? No. Even with a generous ~60% reduction in crash rates per million miles driven between normal and L3 autonomous driving, we’re still looking at 21 fatalities per 100,000 hours driven. Try to ignore this when taking your next Uber.

But what about Kobe Bryant?

We all fall into this psychological trap. Kobe Bryant's terrible helicopter tragedy, the recent New York sightseeing helicopter crash - these stick in our memory precisely because they're rare. It's the same reason people fear shark attacks but not bee stings, even though bees kill more people annually.

Consider this: multiple plane crashes kill thousands yearly, but a single helicopter accident makes global headlines for weeks. The rarity makes it memorable, which makes it feel common. It's a cognitive bias, not a statistical reality.

But you have no control in a helicopter

This one's interesting because it reveals our perception of control versus actual control. In a helicopter, you have a wider viewing aperture than any plane - no "metal box with tiny porthole" experience. You can see everything happening around you.

You've got a human pilot making real-time decisions, not an autopilot system. You have higher escape capability than a plane at 30,000 feet. Yet we feel more "in control" in a self-driving car or on a train where we literally control nothing.

The control argument falls apart when you realize you have about the same level of control in helicopters as in trains, planes, or increasingly, cars.

The Real Question

The real question isn't why helicopters are scary. The real question is why are we bad at risk and value assessment? We reject the higher cost option due to poor value assessment. We fear the statistically safer option because it feels unfamiliar, while accepting higher risks in our daily commute because they feel normal.

Next time you're stuck in traffic for 120 minutes to get somewhere a helicopter could take you in 10, ask yourself: what are you really afraid of?

Subscribe

I’ve always written on Interspace because I’m a fintech nerd and enjoy writing. What started as a way to learn to write compelling fintech Twitter threads (and recover from the burnout of my hyper-growth startup career) became one of my favorite interests. Making the complex simple and reconstructing thoughts to fit single 280-character tweets that engage someone to page through 10+ tweets at a time was incredibly captivating.

As my posts became more frequent, Twitter’s algorithm favored shorter takes or singular ultra-long tweets. With worsening search features, I moved my over-engineered fintech tweets to Substack to become over-engineered fintech posts.

The format worked better than Twitter threads and allowed me to expand my thinking to longer essays with more data, charts, links, and multimedia. I found it easier to broaden my writing interest from fintech to AI, governance, and technology. Eventually, Substack’s limited features and theming became a blocker, so I moved my newsletter from Substack to Beehiiv.

For customization, brand equity, and product innovation, it’s the clear choice for any newsletter creator. Beehiiv is better in almost every way except for Substack’s stubborn network effect around its recommendations feature that generates subscriptions. Beehiiv will get there eventually as it launched Top 4 to provide the Recommendations network effect, and it’s better in every other way, but it’s worth noting Substack’s lone standout feature.

Moving to Beehiiv prompted me to think about monetization and rethink my approach. I didn’t realize this at first. As a full-time operator and because writing has only been an interest, I never considered monetization. But money is a strong incentive for focus in a busy schedule. When folks like Sheel Mohnot make ~$50K / year for posting interesting content (alongside a full-time VC job) and CJ Gustafson left a lucrative career as a high-powered CFO to run the Mostly Metrics newsletter with ~200,000 subscribers, I wondered about the potential.

Broadly speaking, there are two main ways to monetize attention:

1. Subscriptions: Your users pay a recurring fee for access to premium content. You can have a fully paywalled experience or offer some free content with subscriptions for ongoing or exclusive access (Mostly Metrics, Fintech Business Weekly, Payments Culture)

2. Advertising: Your users see advertising, promotions, affiliate marketing, and other forms of advertising in your newsletter. Advertisers pay you for the audience and/or engagement (Fintech Takes, Popular Fintech, Fintech Brainfood)

Share this post from Interspace

Subscriptions

The playbook for Option #1 is simple. Get as many people to read and subscribe to your content and price it to maximize subscribers and optimize revenue. We can derive key unit economics from Lenny Rachitsky of Lenny’s Newsletter. In under 5 years, he went from 0 to 1.1M subscribers. His paid subscription is ~$10 / month (often a corporate L&D expense). With 15,000 free subscribers and 500 paid members (~3.3% conversion), he was netting ~$65K/year before fees (Substack takes ~20%). At 1M, he should be netting ~$4-$5M / year in gross fees. That’s a lot of money.

I now have about 1,000 subscribers. With a conversion rate of ~3%, I’ll have 30 paid subscribers at $10 / month, resulting in about $300 per month if I started now. It’s a start. To do subscriptions well, you must gate information between the free and paid newsletters to make paying feel worthwhile. I’ve seen folks limit the number of posts for free vs. paid, provide special content for paid subscribers only, and/or exclusive events for subscribers. You don’t need everything at once, but there needs to be some gating or your free subscribers won’t convert.

That sounds like a lot of work, and it interferes with what I started this newsletter to do: inform and distill complex concepts into digestible essays.

I don’t want an entrance fee for information I’m trying to make accessible.

Advertising

With advertising, the approach is different. Advertisers pay for attention and engagement, so the goal is less about convincing an engaged audience to pay for information access. Instead, the goal is to build a large or engaged memberbase worth spending money to promote products to.

Generic advertising

Advertisers will pay per impression or qualified engagement, with CPM (cost per 1,000 units) as the primary metric and the units being measured as views or engagements. Values range between $5 - $10 per 1K impressions. Twitter recently started monetizing with CPMs at ~$4 per 1,000 qualified impressions (engagements: likes, replies, reposts). Twitter engagement is lighter than a blog post, so we will use a YouTube or Pinterest CPM instead. If I ran generic ads on my blog posting ~2-3x / week at ~$15 per 1K engagements, I will earn ~$180 per month.

That said, I don’t want to charge for information access or dilute my content quality with generic ads. So this isn’t a realistic option for me.

Sponsored Advertising

One way to do this is to spend more time with brands and advocate for them as a product user (paid influencer) or through sponsored posts. Given the Interspace community comprises operators from major fintechs and exciting startups, I could work with brands to craft narratives that explore topics interesting to readers and appealing to brands.

How much could that accrue for my newsletter? Packy McCormick, of Not Boring, charged ~$5K per post and ~$20K for a sponsored deep dive when he had ~36,000 subscribers ($140 per 1K subscribers). I could charge at least ~$150 per post and $500 for a deep dive. Adam Ryan of Workweek suggests Packy might charge $200 CPM or ~$160 post-fees. At 2-3 posts per week and ~1 deep dive per month, that might be $2,000 monthly.

I would need to maintain my post frequency and ensure my brand partnerships resonate with Interspacers. I would need to find the right partners who believe in our vision, and have a network of fintech operators to choose from.

But the incentive alignment is there, the content can remain free, and there can be a win-win when I am sponsored by products I use, people and thought leaders I respect and services I trust.

This is the way.

Interspace is going to the moon

With that in mind, Interspace is going sponsored over the next few months and some future content may be sponsored with explicit sponsorship tags and guidelines.

As always, content will remain now and forever free. And while sponsors may pay for my posts, editor ownership will always remain with me; brands choose the types of content they want to sponsor but will never get direct editing privileges.

Ultimately, I want to align incentives and ensure writing stays a focus of mine when other priorities in life become increasingly in demand of attention. Help me make Interspace the priority because I have so much more to share.

If you want to reach the leading minds in fintech, stratfin, tech, AI and politics, respond to this email and we’ll make it happen.

Coinbase is building a modern payments network from scratch, using the best traditional and blockchain infrastructure. The future of fintech payments is here and we are witnessing something transformative.

I didn't connect it until Jev explained it clearly:

1. Stablecoins addressed the volatility problem.

2. Coinbase’s Base addressed the speed and cost problem.

3. Coinbase Commerce protocol can now address the commerce transactions problem.

The missing element is chargebacks and representment, and I can't figure out why they skipped it. It's crucial to commerce and I don't see a replacement, as it ideally funds the disputes system which increases trust between merchant and user and should boost acceptance.

Kazanins and many others believed the system was intentionally skipped because Coinbase didn’t view it as necessary for a modern payments protocol. “It’s a cool feature for consumers, but it harms merchants (and is likely abused by consumers).” Based on their Shopify experience, the fraud and uncertainty from chargebacks didn’t justify the user trust benefit or offset the operational and financial expense of maintaining this dispute system.

Coinbase seemed to share the view, but I wonder if it was biased by their design partner, Shopify, dedicated to serving merchants. Matt Jones of Payments Culture pointed out that “no one wants to build chargebacks because it’s really hard” but as a user, if you can’t chargeback a large purchase like travel expenses, you won’t make the purchase. This was the sentiment of many in the commerce space, that chargebacks are essential for users to adopt the new payments network.

With all that said, there is no issue at all if you think about where the responsibility for maintaining a disputes & chargebacks system lies. Kazanins noted that the protocol doesn’t prevent Operators from building a chargeback mechanism on top of it, giving Coinbase the benefit without the responsibility to maintain it. Indeed, Jones pointed out the difference between rails (the protocols in which payments are made) and schemes/networks (the methods by which stakeholders agree to transfer payment data between each other). It is the schemes, like Visa and Mastercard, that have historically developed and maintained chargebacks & disputes systems, with each having their own procedures and infrastructure.

Coinbase is taking an explicit stance that chargebacks should happen at the schemes level versus the rails level in their omission of the chargebacks & disputes system at the protocol level. They should have mentioned or guided this concept to get buy-in from traditional ecommerce participants wary of a new protocol.

The protocol’s introduction may help us challenge our legacy assumptions around chargebacks & disputes. Should we remove the entire system since it could be an unsustainable and expensive one that doesn’t support itself as some argue? Or should we take a new approach with different chargeback options for different risk merchants with onchain transactions as a deterrent and built-in incentives to price and manage risk as Jones suggests?

Ultimately, as Jones says, “If someone can build a solution where the competence and value assigned by issuers isn’t a factor in dispute success, that would be a net benefit of any new system.” This is a future I am excited to see.

I’ve seen more focus on earned wage access, but fintech nerds may not understand the opportunity or unit economics. So, it’s time for another over-engineered deep dive.

To set this up, let’s define Earned Wage Access / Early Wage Access / Earned Wage Advance (“EWA”). At its core, EWA relies on three tenets:

1. Specifically for intra-paycheck liquidity

2. Extremely short duration (<2 weeks)

3. Non-recourse

These tenets define EWA’s core principles and minimum requirements. Non-recourse is a critical element, allowing for the key features of the typical EWA product:

• No fixed repayment schedule and limited ability to collect on outstanding advance beyond the agreed-upon revenue stream (the specific paycheck).

• No need for state lending licenses for originations.

• No requirement to disclose the APR or structuring fees as APRs, which would make these products seem predatory.

Many EWA flavors exist, but they vary across two dimensions impacting risk and core business model and unit economics: 

Direct Employer Connected vs. Cash Flow Underwritten.

Direct Employer Connected (“DEC” EWA)

With direct employer-connected EWA firms, EWA is provided via a direct connection to the employer’s payroll. This allows sophisticated payroll & time tracking to calculate earnings and provide an advance.

Key benefits & Considerations:

(+) Lower underwriting risk due to verified calculated earnings

(+) Can offer services for free and monetize through card interchange and/or instant transfer fees.

(-) Complex tech stack requiring multiple integrations with large payroll providers

(-) Lengthy sales cycles to acquire employer connections

The structure makes it difficult to run DEC EWA profitably without major scale because the upfront tech investment and long sales cycles mean slow but consistent growth. However, strong unit economics allow successful players to become profitable quickly with the right distribution.

Unit Economics (for $200 wage advance):

+$3.00 fee charged to payroll provider / employer for EWA

+$0.50 for customer service & platform (charged as platform fee and amortized)

-$2.00 estimated losses (1% loss rate)

= $1.50 (75% contribution margin, excluding financing)

Key Players / Competitive Landscape (by estimated valuation and target employer size)

XL (>$500M): DailyPay (Enterprise), PayActive (Mid-size)

L ($150M - $500M): Clair (SMB), Rain (Enterprise)

M ($50M - $150M): 

S: (<$50M): Clockout, TapCheck, Wagestream

M&A / Exits Landscape

M&A can be limited because small players are generally sub-scale as they build employer integrations and quickly become scaled with no easy M&A entry point. Once key employer integrations are set, it takes time and/or select partnerships to accelerate growth.

Conversely, only with DEC EWA do you see significant independent players because unit economics are favorable, growth is steady and low cost, and profitability doesn’t depend on building out the product suite.

Cash Flow Underwritten (“CFU” EWA)

With EWA, firms using cash flow underwriting use sophisticated transaction analysis to estimate earnings and provide an advance.

Key benefits & Considerations:

(+) Faster sales cycle since it can be offered directly to customers.

(+) Simplified tech stack with no need for direct employer integration.

(+) Can run a business profitably at small scale.

(-) Higher underwriting risk due to EWA loss rates (CFPB estimates ~6%)

(-) Need to charge fees to counter underwriting risks.

Unit Economics (for $200 wage advance):

+$5.00 user access fee

+$2.00 Instant Transfer fee

-$0.50 Instant Transfer expenses

-$3.00 Loss Rate

= $3.50 50% contribution margin (ex. financing)

Key Players / Competitive Landscape (by estimated valuation)

XL (>$500M): Chime, Earnin, Dave, MoneyLion (acq. by Gen), Truebill (acq. by Rocket Money)

L ($150M - $500M): Brigit (acq. by Upbound)

M ($50M -$150M): FloatMe

S: (<$50M): 

M&A / Exits Landscape

There’s a clear opportunity to acquire sub-scale or ramping players with ~$50M in contribution profits, and there is historical precedent for this. Valuations offset significant growth rates against consumer target demo (vs. B2B view of DEC EWA).

On the flip side, it’s rare to see scaled independent CFU EWA players because unit economic margins are tight, customer acquisition is expensive, and profitability often requires building out the product suite or achieving mega-scale.

Of course, it wasn’t clear that my Chime IPO prediction would come true. On April 1st, Tariff Day forced several fintechs to delay their IPO plans. That day, I claimed that rather than a permanent halting of their IPO, we’d see a delay until Q3 or Q4 2025, especially if markets improved and tariff actions were reversed. That’s because Chime’s financial performance is outsized in Q1 2025 as their users get and spend tax refunds. A May IPO would allow them to report only full year 2024 financials on their S1 and have their first reported quarter be a strong Q1. This tried-and-true strategy was the same one we used in the majority of fundraising rounds when Chime was a hyper-growth private company.

By May 8th, despite a stock market rebound and the administration’s efforts to walk back the tariffs globally, it didn’t seem like the administration would be able to counter China’s asks. Still, I argued that given some Q1 tax refund spend had shifted to Q2 and MyPay growth could offset the normal Q2 revenue slump, Chime might file within 1-2 months with an IPO by Aug-25 at the latest.

After the surprise trade deal announced on May 11th, I remarked on May 13th that Chime could file its IPO docs publicly by May 14th, 2025 without updating its financials from the original registration statement, just two days later.

On the same day, Chime filed its S1 publicly, setting the stage for a historic IPO in the window I set up with an IPO date from May to August 2025. As predicted, they filed with 2024 audited financials but provided an unaudited look at another strong Q1-25, likely because they expect Q2-25 to be even higher so they’ll have two great reported quarters, just as I had thought.

There’s a lot to digest in the Chime S1 and I’m reserving my thoughts for later when the dust settles to add to the discourse. There are already some great takes from fintech nerds that I’d check out:

• Nick Madrid, a former Chimer from The Free Toaster, comments on Chime’s ARPAM flywheel.

• Zu Daya, from Strategic Finance Careers, discusses the metrics and proposes a potential valuation range.

• Simon Taylor from Fintech Brainfood explains the significance of owning the primary financial relationship. 

• CJ Gustafson’s Mostly Metrics runs the most detailed S1 breakdown available, supported by the Run The Numbers interview of Chime CFO Matt Newcomb.

• Jason Mikula from Fintech Business Weekly covers their marketing metrics and provides another view on valuation.

For now, here’s my over-engineered take on Chime’s Q2-25 performance: 

We know that IRS filings delayed tax refunds that normally fall in Feb/Mar to April. We saw that CashApp’s recent shareholder letter noted tariff uncertainty led to a Q1 spend pullback shifting to April. As a result, we’ll see spillover spend boosting Q2-25 interchange revenue.

We also know that Dave and other companies with short-term lending products outperformed in Q1-25, so MyPay will benefit from the same tailwinds. MyPay will dominate the topline growth story (but struggle on the bottom line), boosted by new member acquisition from the IPO brand play of a household consumer brand.

Q2-25 should thus give us strong topline revenue, great new member acquisition but perhaps a narrowing operating profit margin as the relatively lower margin MyPay offsets the otherwise SaaS-like interchange profit margins.

Let’s see if my fintuition holds true for Chime earnings and beyond.

The trade war changed a lot of things in the fintech ecosystem, but potentially none more jarringly than the prospects for IPO for major companies like Circle, Klarna and Chime. A few weeks ago I made the claim that Chime's IPO would be delayed, but not necessarily halted and that if the stock market improved there would be a chance that Chime could still file the S1 by the end of the month with a target IPO before mid May 2025 (see original view at end of the post). This is no longer possible.

My predictions at the time (one day after "Tariff Day") were:

1. America will pull back and renegotiate tariffs (correct)

2. Stock market will rebound (correct)

3. It would be in Chime's interest to file S1 with 2024 financials so they could report a mega Q1-25 on the back of tax refunds season (correct)

4. It would in Chime's best interest to file S1 in near-term so their first reported quarter would be larger than any S1 filed report, avoiding 2+ quarter delay (mostly correct)

So if all of those predictions were correct, why didn't Chime file their S1? What I now think I didn't get fully correct was exactly how (4) would play out. I still think that Chime is delaying their IPO not halting it, and there's a good chance that it's only 1 quarter.

Here's why:

A. The gutting of the IRS significantly delayed tax refund timing, so a portion of tax refunds normally spent in Q1-25 will now be spent in Q2-25

B. Even though most tariff policy was pulled back, it created lots of uncertainty and volatility that reduced consumer spend in back half of Mar-25

C. Chime's installment loan and MyPay product likely growing quickly and can offset seasonal decrease in Q2 revenue vs. Q1 revenue at least in near term

D. Chime has built up enough brand equity, hype and investor expectation around a 2025 IPO that it won't be prudent to wait longer

What does this practically mean? There's enough indication for Chime that Q2-25 is going to be better than Q1-25 that they are okay with waiting out the volatility and regrouping around a short-term delay. If volatility does subside, we should see an S1 filing in the next 1-2 months with anticipated IPO by Aug-25.

Original thinking circa April 1st 2025

My completely personal view as an experienced fintech operator who understands the ecosystem well on what is happening with Chime's IPO. It does not represent the views of my current or former employers - it's my own over-engineered fintech take.

The Klarna IPO delay is definitely not a great indicator for Chime IPO timing (in the next 4 weeks). If Klarna couldn't get enough commitments to be comfortable with IPO at the price they've set, there's no way Chime is going to follow suit (yet). I had previously thought they would drop an S1 right around today with a target IPO date of the last week of April. I no longer think that's the case.

That said, it's not all tears and fears. There's some indication that the pain we're feeling now in the stock market will improve over the next few weeks. The US literally put a tariff on everyone, how much worse can it get (famous last words). More likely, America pulls back or renegotiates some of these tariffs.

And we can't argue with the fact that at current prices, any stock you pick is definitely on sale. If you have any shred of optimism, you're buying stock now. It's no wonder that retail investment inflows was the highest it was in the past decade. And it's not surprising that even in this market, Circle filed its S1.

I also think that practically speaking, Chime will want to IPO before Q1-25 financials are due. Unless markets get materially worse, there is no way they want to have to rewrite their S1 and talking points by letting this delay more than a few weeks. We've seen Chime launch Chime+, mention Instant Loans, announce Chime Enterprise, wrap ATMs. They built up a TON of brand around a now or soon to be released S-1. That's an important cost.

That's in addition to a lot of business / financial reasons for Chime to launch in early Q2 and it's all about the fundamentals of the business. Chime's core demo is paycheckers, and this demo tends to get tax refunds and spend it. That means Chime's Q1 is going to be massive, as it always has been. There's a reason we ran 80% of Chime's fundraising rounds around Q4/early Q1. It's a proven formula.

Chime will want to have an S1 with full year financials, IPO, and then immediately report Q1 which will be outsized. They do not want to report Q1 in S1 and then have to report a lower Q2 after. It feels weird to say this about fintech, but for this demo, the business IS seasonal.

So for Chime, the decision tree is - do we do this IPO now (or in a few weeks) or do we push this out at least 6 months or longer? I definitely can see a small delay, but the idea of pushing this out 6+ months for some future uncertain IPO window seems pretty unrealistic. I can't see them doing that, knowing what we know today.

Where does that leave us? I think Chime will still post their S1 sometime in the next few weeks with a target IPO date of no later than end of May. Any time after that and the SEC will pressure them to include Q1 financials which just throws off everything.

Let's see what happens!

Subscribe

If you know me, you know I’m a huge Duke fan. My first year was when Jon Scheyer captained the national championship-winning squad in 2010. So March Madness is a big deal for me. I watch as many matchups as I can while filling out a bracket usually ruined by the end of the round of 64 (which needs a better name like “Big 64” or “Nintendo 64”).

When Robinhood announced they would work with Kalshi and DerivativesX to surface March Madness prediction markets in-app, I couldn’t believe my fintech and sports worlds were colliding and I had to check it out.

I’ve always been interested in sports betting as a non-correlated supplement to stock market investing, but I found the practice arcane and uninviting, often requiring sophisticated knowledge of unintuitive terms and access to questionable gambling websites that only accept cryptocurrency. Robinhood’s promise to democratize access and simplify the market like they did for stocks, options, and crypto resonated with me, and I knew that, more than beating the odds, it would be a good way to learn how fintechs could innovate on and improve a legacy practice.

Some people think Robinhood shouldn’t enter sports betting, but they are making broad generalizations and big assumptions that hinder their ability to think through their reservations. First, we should define what we’re arguing about and ensure the legal foundations are clear.

When we talk about betting on sports, there are two major ways to do this:

1. Sportsbook-based sports betting (“traditional” sports betting): A centralized entity (the bookmaker) determines the odds for a sports event and provides pricing (the spread) for the outcomes, inclusive of their fee. Bettors place bets against the bookmaker, who balances the spread between bettors on both sides to pay winners.

   1. Sportsbooks’ business model relies on determining the spread and optimizing based on real event odds, as it dictates the bookmaker’s revenue (e.g. DraftKings, FanDuel).

   2. Since the bookmaker sets the odds and payouts, there’s no federal regulatory body governing these entities. This leaves legality and enforcement to the states.

2. Exchange-based sports betting (“new age” sports betting or “sports trading” or “sports prediction markets”): An exchange provides a platform where bettors can bet directly against each other, rather than against a centralized bookmaker. Sports event contracts are largely binary (win/lose) rather than based on point differentials, initial odds are set at 1:1, and the market determines the spread through buying and selling event contracts with a small per contract trading fee. After the event, the winners take money from the losers.

   1. The exchange-based sports betting model depends heavily on platform trading activity. The exchange doesn’t care who wins or loses, only that many people are trading (e.g. Kalshi, DerivativesX, Robinhood).

   2. Since the exchange is a platform for facilitating trades, federal regulatory bodies like the CFTC govern these entities and provide legality and enforcement at a federal level.

In short, exchange-based sports betting:

• operates on a federally regulated, open platform where odds are clear via the market order book.

• allows trading against other bettors instead of a centralized bookmaker, reducing incentives to lower bettor odds.

• allows live, real-time unlocked bets that can be closed out or added to at any time during an event

Now, in order to engage in this debate, we need to understand Robinhood’s position and the regulations. Sportsbook-based sports betting is federally allowed and legal in 40+ states, and the practice will continue regardless of Robinhood’s participation. Traditional sportsbook-based sports betting is associated with negative perceptions of gambling including dark patterns that encourage high-risk plays like parlays, adversarial odds-setting, and access cutoffs for players that “beat the house.” This type of sports betting can have negative expected value when spreads are determined in a black box way and bettors on both sides interact with a bookmaker, not each other.

But despite being legal in many states, Robinhood isn’t even operating in the sportsbook-based sports betting space. Robinhood, Kalshi, and DerivativesX are in the exchange-based sports betting space. In this space, bettors trade directly against each other, and the expected value is zero, not negative (other than the nominal fees). The fees are transparent, and it’s in a platform’s interest to make the spreads as clear and real-time as possible to encourage trading. Robinhood is actually one step removed in that it surfaces the contracts that Kalshi/DerivativesX provide in a user-accessible way. Stopping Robinhood from accessing exchange-based sports betting only stops good UI, not the practice itself.

There are at least five arguments against sports betting:

1. You think sports betting should be illegal and fintechs shouldn’t be able to enable it

If you have an issue with fintechs engaging in sports betting, talk to your legislator to change the legislation. Don’t stop companies from operating legally. You should also justify why we legalize vices like cannabis, alcohol, and smoking but want to crack down on gambling. Explain why banning sports betting will lead to better outcomes than Prohibition or other attempts to ban vice versus drive incentive alignment via penalties.

I don’t like profiting off vice, but let’s not apply rules to one versus another or claim to know what’s best for Americans. Evidence shows sports betting can engage consumers and have an economic impact (over $300bn annually). Despite the charged topic, 80% of US states have legalized it. The people voted for this - why should you determine its enablement?

2. You think sports betting should remain as it is because it’s efficient and well regulated

Are we discussing the same industry where two major players face hundreds of class action lawsuits and have settled for millions? You’re defending an industry where centralized sportsbooks operate an oligopoly with FanDuel and DraftKings controlling 70% of the legal market and the top four players comprise 90% as somehow “good” for society? Or are you satisfied with the fact that despite progress on online betting, 60% of all sports betting occurs on unlicensed platforms with no consumer protections and high exploitation risk? I hope that’s not the position you chose to defend.

Source: AmericanGaming.org

3. You think that sports betting is more dangerous and less “good for society”

This is patronizing and inconsistent. How are trading memecoins, 3x leveraged tech shares, zero day options, bitcoin ETFs, and index options “better for society” than sports betting? If you believe this, I have some SPACs, penny stocks, and dividend arbs to sell you.

Some may argue that holding stock is more long-term than sports betting. I would point them to the rise in day-trading, memecoin investing, and the declining average hold duration. Why hold onto a stock when we have real-time information on its trend and finfluencers claiming to know our investment moves? I would also ask you to justify why people invest in financial engineering and price improvement like the market-makers driving 80%+ of market volumes.

4. You think banks and fintechs should only provide services for financial health

Have you ever shopped for financial services in America? Alongside Robinhood’s prediction markets are high APR cash advances, payday loans, memecoins, hyperyield-bearing stablecoins, high fee earned wage access products, shady alternatives, complex real estate investments, and stock market derivatives, all available at a bank or brokerage. Robinhood should not be the only one above your line while Schwab, Fidelity, and E*Trade are saved from ire. And maybe go after those big banks that provide for financial wellness but have caused $20B in overdraft fees, predatory lending, and reductions in financial access.

5. You think sports betting is too risky, so we should ban it outright or prevent fintechs from having access

I find it hard to believe that sports betting is riskier than other investments allowed for regular investors and enabled by fintechs. Real estate trusts, home flipping, crypto, venture investments, hedge funds, infrastructure funds, etc. Why do we draw the line at sports betting when many are riskier, especially on an exchange with binary outcomes, making the chances 50/50 for most bets?

If we have issues with regular retail investors adding “sports betting” to their portfolio, why not use the existing investor accreditation feature instead of banning or limiting fintech access? There’s no reasonable argument to bar a savvy investor from investing in legal sports betting.

The power of prediction markets

The backlash against Robinhood stems from a misunderstanding of exchange-based vs. sportsbook-based sports betting. The claims that Robinhood isn’t “democratizing finance” or seeking truth are also incorrect. By allowing more regular investors access to the sports betting market, it democratizes finance (reducing fees and increasing transparency).

But more profoundly, having bettors determine the markets ourselves is crowdsourcing and amplifying intelligence gathering, increasing our ability to pursue truth. Dismissing prediction markets’ power due to the near-term sports applications misses the point.

Instead of waiting for closed-door meetings and filtered public disclosures to understand interest rates, we can utilize the hive mind to gauge expectations in real-time. We will no longer be limited to quarterly government reports rife with errors or talking points from politically influenced leaders; we can go to prediction markets for the truth. If you can’t grasp prediction markets, look at the stock market’s real-time access to a company’s performance compared to their one-sided shareholder letters and quarterly earnings reports.

I’m excited about prediction markets, not just for March Madness sports betting (forget stale brackets for real-time betting each round) but for all national financial KPIs and international metrics.

We’re putting our money where our mouth is, and that’s a good thing.

If you haven’t heard potential future presidential candidate Pete Buttigieg go straight into the lion’s den to debate his critics on the Flagrant podcast, start watching. Among the many engaging exchanges was one about government research.

I largely agree with @PeteButtigieg's take (similar to @hasanthehun and @BernieSanders) on principle, except the idea that only the government can invest in long-term return ideas. In fact:

1. The government doesn’t necessarily have a special long-term view on the initial creation of its large-scale research projects (it is an open question on whether the internet was ever meant for public usage)

2. The government isn’t the only entity that can come up with multi-trillion dollar ideas.

We are experiencing a multi-trillion idea right now - conversational AI. It wasn't developed in a government lab; it was created by private enterprise (@OpenAI). The notion that private enterprise can only be short-term focused is literally short-sighted.

Government should play an important but indirect role. Non-profits and governments aren't effective at driving scalable change because they're not focused on sustainability (i.e. profit). Combined with a near requirement to sacrifice pay & benefits "for the greater good" to hold an American government position, it makes it very difficult to retain the best and brightest.

Here's how the right government should operate:

1. It should provide consistent regulatory frameworks that counter capitalism's tendency towards monopoly and negative outcomes from unrestricted capitalism. With constituency support, it will direct funding and regulatory favorability towards broad causes like stopping global warming or rebuilding American infrastructure.

2. It should raise and direct funds to public goods with challenging business models, but only through the funding and support of non-government enterprises.

3. It should provide an enforcement layer for companies violating regulations and a national defense system. It should have ultimate power and ability to enforce regulations approved by its constituency.

That's it.

Of course, there are important implications:

(A) We need to overhaul our entitlement programs, not to eliminate them but to install private enterprise leads that care about balancing budgets, optimizing efficacy, and driving social change.

(B) We need to increase government power to regulate bad actors. When the Mag7 or big corporates violate regulations, penalties must be strict, substantial, and swift.

(C) We need to rethink or dismantle the government apparatus managing the IRS, NIH, TSA, CBP, DEA, SSA, DOT, Education, and even the USPS among other agencies.

In practice, government entitlement programs and research labs will transition ownership to private enterprise over time, though the government may still provide significant funding. Government programs like IRS and USPS should enforce regulations around free and fast tax filings and mail delivery rather than attempting (poorly) to provide these services directly. Government enforcement agencies can remain but shouldn’t enact independent programs other than those that enforce regulations.

People on both sides will immediately challenge (C) and some will resort to ad hominem arguments and/or out-of-context misreads to counter this position. Some may say these changes are unprecedented and anarchic.

But for every example I've provided, I can point to a developed country that does what I'm advocating for and does it well:

- Privatized transportation: Consider Japan's subway infrastructure or the Public-Private Partnerships shaping many countries' infrastructure development.

- Privatized research: We already operate this way in many respects, and NASA shows how funding private enterprises instead of building directly has led to improved outcomes.

- Privatized healthcare: Switzerland and Norway require private insurance, and their governments heavily regulate it to ensure universal healthcare.

- Privatized education: I attended public schools and want my future kids to go to one. However, American education needs a major overhaul. Public charter schools, funded partly by the government but privately operated, exist in the US and should be more widespread given their performance.

- Privatized taxes: In most countries, employers/banks drive tax calculations because tax authorities pre-fill returns based on this data for user approval instead of requiring the user to fill it out.

- Privatized utilities: Look at the UK and France for examples of a largely privatized utility infrastructure that works and is efficient. Large-scale energy markets establish consistent pricing across markets.

- Privatized logistics: Look at Germany and the UK, where privatized enterprises operate under government regulation.

I share a common cause with @AndrewYang, @pmqrcq, @garrytan, @ezraklein and others who believe that (1) tech can support abundance as the primary driver and (2) government should regulate and enforce right action.

What do you think?

Here's my hot take regarding OpenAl's flop with 4.5:

First, they didn't want to lose the spotlight. Plain and simple, 4.5 was clearly half baked and with Grok and Sonnet coming out, Sam Altman didn't want to be seen flatfooted.

Only problem is it backfired.

Now, here's my deeper analysis:

Ilya left OpenAl in May 2024, but was part of the ouster attempt in November 2023. All the progress that OpenAl has made since then was primarily the momentum championed by Ilya. In other words, Q* and Strawberry and all that were llya, and Sam made the miscalculation that he could continue research at the previous pace without Ilya. But then Mira and a ton of other researchers jumped ship because of the profit maximizing strategy that Sam took, throwing caution to the wind. I mean, OpenAl reneged on literally every single one of their founding principles. For profit, closed source, reduced focus on safety.

"We don't have enough GPUs"

That could be part of it. But also consider that Sam's pedigree from Y Combinator is rapid startup, rapid exit. OpenAl is no longer in the startup phase, they are in the hypergrowth and sustaining growth phase. Sam is completely out of his depth with respect to the phase of the business. Sam Altman has never run, grown, or scaled a company as large as OpenAl. This is why they have a scattered product strategy and focus more on hype than substance lately.

It's actually pretty common for companies around the size of OpenAl to have a total change of leadership. This is often when the exit would happen. The founding team, having built something that is clearly resonating, but not yet profitable, would sell out to a more mature, experienced management team who then bring the product strategy and margins to the next level.

Correct me if I'm wrong, but Sam has no experience actually making a company profitable.

  David Shapiro @DaveShapi

And it got me thinking:

• (A) Could Sam’s position ever be not secure?

• (B) If he ever decides to leave, who would be the right CEO for him and for the company?

On the topic of (A) after what happened in 2023, my thinking is there is absolutely no way Sam is no longer CEO unless it's on his own terms. He cannot and will not be pushed out. Who leads OpenAI is completely up to Sam.

But on (B) I think there's a shot that he recognizes what he's good at and that to take the company to the next level and realize the full value of his equity, he needs to stand aside.

But he's not going to stand aside for just anyone. He needs that person to have a specific and special set of skills and experience:

1. Understands R&D-led companies with developer-led GTM, enterprise sales and API-driven business models

2. Well respected in startup ecosystem with connections to Y Combinator

3. Seen company growth from idea stage to thousands of employees

4. Driven a founder-led business to IPO and beyond

5. Successfully navigated complexity of dealing with multiple global stakeholders and investors

6. Comfortable with build/buy/partner approach to company building with a history of successful acquisitions and partnerships

Who is a person that has all of these characteristics and happens to available to jump into this multi-billion dollar company? Who could possibly be ready for all that OpenAI requires?

That person is Jeff Lawson, former co-founder and CEO of Twilio.

He fits every quality, characteristic and life experience necessary to lead OpenAI to its next stage. He's the person that should take on the role. Sam - you are so great at building companies and you need to stay at OpenAI to build. Carve out a President role and lead OpenAI's moonshot factory. The future is already here because of you so help lead OpenAI to what's next.

I got an iPhone 16 Pro and a Macbook Air and was ready to go. Unfortunately (or fortunately?) I quickly found that on some levels, the promise wasn’t the reality.

Could iOS be the operating system for “design guys”?

I wasn’t switching to iOS for anything to do with messaging. For better or worse, as an Android user, you’re not even conscious of the difference until your iOS friends remind you of their broken chats (the blue bubbles are completely caused by Apple’s restrictive cross-platform messaging stance).

I originally intended to switch for completely different reasons - for the apps themselves. While people claim that at this point, both operating systems are so mature that the available apps should be identical across platforms. While that’s true in theory and for the top 25 apps, it’s not so for a number of specific apps for tech nerds like me. And after years of frustration in the Android Play Store, I wanted to experience something better.

One note for the audience - I’m what my brother refers to as a “design guy.” I care a lot about the aesthetics of an app and to me it’s as important and sometimes more important than function. In the marginal tradeoff between form and function, after a certain point of minimum viability, I will trade form over function. Because for me, form is function and a well-designed app says a lot about the developer’s focus, attention to detail and care for the user experience. 

With Apple, you have to pay for quality

So iOS already appealed to me for those reasons, but I had found a way to mostly get Android to conform to my design principles, at least in terms of the system settings, launchers and app config. But apps were another story and I was reliant on developers to share my point of view. While there are diamonds in the rough, it’s unfortunately abundantly clear that Android apps focus on function whereas Apple’s balance form and function.

Here were the apps I was ready to switch my mobile operating system for:

1. Copilot Money ($13 / month): Based on the screenshots and gushing reviews, I felt like I needed this money tracking app. I’ve used Mint, Rocket Money and a whole host of other tools but never found my default app

2. Fantastical ($5 / month): I use some version of my calendar app every day, maybe every few hours and being able to use natural language processing, smart event openings/bookings, all in a design-focused package with better offline support felt so right

3. Sequel ($3 / month): I am constantly keeping lists of everything I want to see, play, read and listen to - it’s a daily activity. So making that activity media rich and delightful really appealed to me over noting things down in Google Keep

4. Flighty / Tripsy ($4 - $10 / month): We had a crazy year of travel in 2024 with an average of two flights per month. I knew I needed better trip tracking and didn’t want to be stuck with the seemingly anti-design app TripIt in the Play Store

5. Bear Notes ($3 / month): In what will likely become a post later this year, I have been on a lifelong journey to find a better notetaker and Bear seemed like the happy design-focused medium between simple but useless Google Keep and useful but complicated Obsidian

And for the most part, I was able to find equivalent alternatives in the Android Play Store with some dedicated searching.

1. Rocket Money (Free): There is no question that post acquisition of Truebill, this is the single best way to reliably track spending, investments and all manner of linked accounts and of course will help you with keeping your recurring spending on track. Interface is just as clean and consistent as any other app and I found navigation a lot better than Copilot which felt a little wrapped in its own hype

   1. Upgrade pick: Origin Financial costs $99 / year but is definitely worth it as you’ll get all that Rocket Money offers plus a financial planner, robo-advising and access to a team that’s iterating very quickly

2. Notion Calendar (Free): While I’ve had some reservations about the over-engineering in Notion, Notion Calendar is just as streamlined and design-forward as Fantastical, with all the features all for free

3. Listy (Free): I actually love the simplicity of Listy vs. Sequel and it has all the same features, more list categories and a nice web clipper for full flexibility

4. Wanderlog (Free): Wanderlog has the design chops of Flighty but with the full travel itinerary / trip planning capabilities of Tripsy, all in one package that’s easy to use and better to collaborate with than either of the iOS-only options

5. Notion (Free): While I use Lazy for my note-taking, I think Notion is the most analogous to Bear and with their recent focus on stability and performance combined with a drive towards AI-enabled knowledge management, it’s been surprisingly delightful in recent months.

I also came to realize why there’s such a difference in quality between Android and iOS apps. I had always been confused why developers continued to choose iOS first or iOS only when the Android userbase is 2.5x the iOS market globally and ~60% of the market in the US. But switching over helped me realize what’s up:

• On iOS, nearly 20% of apps are Paid/Premium and there’s an apparent predisposition to pay for quality apps and almost every app that does anything worthwhile has a price or an ongoing subscription

  • Developers are eager to develop in an ecosystem that by default will have paying users

• On Android, there’s almost 2x the number of total apps and ~10% are Paid/Premium, so by and large there is always a free option for an app that you want

  • The proliferation of free apps makes it easy for users to get access to functionality for free but is likely a nightmare for developers who want to put out quality products that cost something to make

iOS is no better than Android and noticeably more restrictive

Meanwhile, I quickly found that iOS was truly no better than Android and noticeably more restrictive in a few ways that were unacceptable when a perfectly good alternative was right there. A few of the things that were needlessly frustrating:

• Back button: No universal back button, for no good reason

• Default apps: No easy way to change default apps for common functions

• Sideloading: No way to download apps outside of App Store

• Notifications: Limited ability to actually swipe notifications left / right (why??)

• File management: Can’t easily access my files or screenshots - file transfer has to be done in limited ways

Apple’s most surprising letdown was its (phone) hardware

When I made the switch, I anticipated having some issues with adapting to a new way of operating but what I thought would be assured was that the quality of the phone in my hand would be better than any Android phone, in the same way that its widely accepted that the Macbook laptop line is the best quality hardware for the price of any laptop manufacturer.

Not so with phones. There’s so much to say here but I think “Created by Ella” handles it best in her recent comparison of the latest flagships: Samsung Galaxy S25 vs. Apple iPhone 16. It’s really not a question anymore that Samsung has outpaced Apple.

Apple’s last hope is MacOS which indeed reigns supreme on both hardware and software (with tweaks)

The hardware of dreams

Now it’s not all bad for Apple because Apple laptops continue to stand out as high-performing, beautiful machines at increasingly competitive price points. Only Microsoft (via its very good Surface series) could potentially compete with the OS + Hardware platform integration that Apple delivers on, but it’s ultimately held back by an operating system that almost by its nature cannot stop its own fragmentation and inconsistent app support. And even then, how do you compete with Tim Cook’s inexorable march to further supply chain integration that results in a Macbook Air M4 with 18+ hours of battery life, razor thin unibody all for <$1K?

An Operating System + App Store that controls just enough for ease of use without getting in the way

What I was pleasantly surprised by about MacOS operating system is that the amount of control it exerts in terms of standardization is just enough without getting in the way (like it does on iOS). I actually do want every application to reliably have a settings page in the topleft menu bar. I do like the standardization and management of notifications so I’m not getting random popups in random parts of my screen. I’m all for the streamlining and standardizing of UIs across MacOS apps and the focus on form alongside function. I appreciate simple app installs and simple app uninstalls - this doesn’t need to be as complicated as it is on Windows.

And man is everything just so fast. You don’t need a top of the line Mac to get top tier performance. It just feels so snappy because of the Apple silicon integration between hardware and software. I didn’t want this to be as true as it is, but it is true.

How to learn to love MacOS with some software tweaks

That isn’t to say that MacOS is perfect. There are still tweaks and issues like any typical tech offering. But what’s great is that there are an abundance of fixes and utilities, so much so that I think the utility ecosystem for MacOS is almost an intentional thing crafted by the Apple design gods to keep users engaged in optimizing.

In any case, here’s my stack of fixes to get Windows-like efficiency within the MacOS package. First off, get yourself SetApp. For $9.99, you get access to 200+ apps, many of which are handpicked as the best-in-class utilities. Think of it like Microsoft PowerToys but instead of relying on unpaid labor of hacker devs, they actually distribute revenue to the app developers themselves. The ROI speaks for itself and it’s MacOS’s best-kept secret.

Here’s the full list of apps I love:

1. AdGuard (free to start): System-level adblocker across all apps

2. Alfred (free): Better app/file/anything search for your Mac

3. AltTab (free): Windows-like app switcher (it's just better, admit it)

4. AppCleaner (free): Simple app uninstall w/ full sys file deletions

5. CleanshotX (on SetApp): Better screenshot app

6. DisplayLink Manager (free): For managing 2 monitors

7. Flux (free): auto color temp adjustor for your computer to avoid eye strain (only needed if using DisplayLink for 2+ monitors)

8. Granola (free to start): Best AI for meetings + summary transcription ever

9. HiddenBar (free) : Organize your menubar (best app name)

10. LeaveMeAlone (on SetApp): Smart unsubscriber

11. Lungo (on SetApp): Keep screen awake (the first of many Sindre Sorhus apps I love)

12. Magnet / Rectangle (free): Better window management & snapping (2/3 1/3 is 🐐 )

13. Menu Bar Calendar (free): Adds an actual monthly calendar to your menubar (Sindre Sorhus can’t be stopped)

14. Novabench (on SetApp): Cross-platform performance monitoring

15. Paste (on SetApp): Super clipboard

16. PopClip (on SetApp): Better context menus for highlighted text

17. Shareful (free): Adds "Save as" and "Save to Downloads" to Apple's Share menu (so so clutch - @sindresorhus you are the 👑 )

18. Spark Mail (on SetApp): Clean, simple email app w/ best compose AI I've used

19. Supercharge (on SetApp): Cut feature and PowerTools for Mac (another sindresorhus gem)

20. Superlist (free): Best todo list by far and it's not even close (the phoenix rising from ashes of Wunderlist)

21. TopNotch (free): "Hides" your notch on your Mac laptop

Waiting for the world to inevitably move to the MacOS + Android meta

I’ve got to say, I’m very excited for the world to eventually come around to the MacOS + Android meta, thereby enabling better cross-platform experiences and better open ecosystems for all. You’re already probably using Android even if you’re not aware of it - Peloton, Roku/Fire/Google TV, Amazon Kindle all run on versions of Android’s operating system. With the continued rise of Samsung’s hardware/software dominance on mobile, it’s only a matter of time before more people recognize the power of Android and build more experiences that are intercompatible across operating systems. It’s already exciting to see some of my favorite modern companies - Superhuman, Superlist, Beeper, Replit, Tidal, Volv build first-class experiences for Android. Meanwhile, the rise of full-featured progressive web apps (PWAs) have enabled developers to build app-like interfaces that work well on mobile and desktop independent of operating system. The future looks bright!

AI's risks are real, but they're met with proactive regulation by global bodies and tech giants' self-regulation. We should not let our deepest fears from anthropomorphizing AI and viewing its rise as a binary conflict misguide us. Ultimately, AI is here to stay and its our duty to enable it responsibly rather than to hold it back.

Leveling the playing field

AI's allure lies in making complex tasks simple through natural language (English is the hottest new programming language), leveling the playing field across socio-economic and professional divides. This empowerment allows non-technical creatives to code and techies without artistic skills to bring their visions to life.

Moreover, in this technological revolution, it’s small companies, not just tech giants, are making significant strides. It isn't FAANG that dominates AI but OpenAI, a plucky startup founded in 2015. And unlike the course of many technologies towards monopoly, AI development seems to encourage fragmentation as specialized models tend to outperform large models on tasks that are relevant for an industry.

Artificial Intelligence is only artificially new

Of course, AI isn't new; it's been evolving since 1956, with milestones like Deep Blue (1997) and AlphaGo (2017) marking its journey. Even ChatGPT's interaction, while innovative, is more a testament to AI's ongoing evolution than a revolution.

Let’s not mistake tangibility with proof of its superior intelligence. Even Sam Altman himself has said that we're at the end, not the beginning of this phase of AI models. So this idea that we're facing a completely new technology seems farfetched.

The misplaced anxiety over job displacement overlooks technology's history of creating as many opportunities as it displaces. The much foretold job displacement (75+ years in the making) has not yet come to pass. The rise of AI might challenge knowledge workers, traditionally seen as immune to tech shifts, but history suggests a positive outcome. New technologies create as many or more opportunities than they displace and AI should be no exception (there are 300 years of history that prove it). 

With all that said, it's important that AI regulation and corporate incentives place value in retraining and upskilling workers so that when displacement happens, there are new opportunities for them. We need concerted effort between governments and the private sector to establish robust educational programs and financial support systems.

With great power has already come great regulation

Meanwhile, the regulatory efforts surrounding AI also surpass those of earlier technologies. Just note the rapid implementation of content moderation policies for every major AI chatbot as a commitment to responsible development across AI models (far more than non-existent early social media policing). Meanwhile, the EU has already enacted landmark AI regulation and the US is not far behind.

At the same time, the biggest tech companies have been proactive about self-regulating AI technology. From red-teaming to ethical study to prominent disclaimers of AI as “experimental” and often incorrect, tech companies are eager to stem the hyperbole (a far cry from Wikipedia’s laissez-faire approach).

What the naysayers also omit is that for the first time in a long time, the very technology we are trying to regulate can be used to regulate itself. Already, governments are using AI to better serve the public and there is no shortage of opportunities to further build out their toolbox.

The only thing to fear is fear of ourselves

Fears of AI's potential harm often stem from a lack of understanding and an overestimation of AI's impending superintelligence and the resulting short-term impacts. But we need look no further than the slow progress in AI-powered autonomous vehicles, once heralded as inevitable by 2020, to illustrate the gap between hype and reality, challenging the immediacy of AGI threats. The idea that we can accurately map AI to the human brain for AGI when we can’t even power AI-driven cars illustrates the arrogance (or ignorance).

Of course, we tend to anthropomorphize AI, attributing human-like motivations and behaviors to machine intelligence, which distorts our perception of its risks. Indeed, incidents of AI-enabled racism or malfeasance highlight the need for ethical training and regulation. However, proper governance can mitigate misuse, encouraging responsible development and application.

In fact, it’s clear that the near-term fear is not AI going rogue, but people going rogue with AI. If anyone is going to stamp out humanity, it's most likely to be humans themselves, using neutral AI technology for nefarious means. The oft-used quote used in the context of AI is that "we have Paleolithic emotions, medieval institutions, and godlike technology" and there are terrible uses of technology that AI that we've seen from human-directed incidents of AI discrimination, misinformation and hacking.

So we should not advocate for the flawed mantra that "AI doesn't kill people, People kill people". But the solution is not to block or dismantle, but to learn from our mistakes and expand our understanding. There are ways we can strictly punish incorrect AI use the way hackers are punished while encouraging AI research to continue.

But by far the most prevailing myth that needs dismantling is the binary standoff we imagine between humanity and AI, as if we're headed for a showdown and we must pick sides. The truth is far more synergistic. Our unique position allows us to bridge the tangible world of atoms with the intangible world of bits, offering us an unparalleled advantage. Today, doctors already use AI to better detect cancer while teachers are using AI to focus more on education than administration. It's not about us versus them; it's about us with them, enhancing our collective potential through augmentation.

We should embrace AI to own our future

Instead of viewing AI as an adversary, we should embrace a synergistic relationship that enhances our capabilities. From healthcare to education, AI is already augmenting human efforts, showcasing a partnership that benefits society.

Our future with AI isn't set towards dystopia; it's a path we're shaping with thoughtful regulation, ethical development, and optimism. By fostering a hybrid approach where AI complements human skills, we ensure technology serves to enhance, not detract from, our collective well-being.

Elon Musk has taken his first pass with an initial goal of $2T in savings by March 2026 which we can all agree would be a huge number worth talking about. If it were the real savings value, but it’s not because that amount is definitely $1T within a few weeks of the end of March 2025. And that number is still quite impressive, but not the real number at all because we all knew all along the number was $150B by 2026. Of course, despite boldly claiming $150B in savings, the oft-cited Wall of Receipts that was supposed to provide evidence for the bold claims still stands at $25B in total savings. If it’s not clear by now, this administration knows neither how to count nor how to apologize for overly aggressive forecasts. In fact, the only thing Elon has reliably cut is his own savings forecasts.

Live look at Elon cutting everything except real federal spending

At one point though, Elon claimed there are at least $55B in savings, which he’d like to remit back to taxpayers in the form of a dividend. So Grok, what's the DOGE dividend going to be? My rough calculations are $55B saved / 350M population = $160 / person ($640 for family of 4). Not bad for a month or two of work. At $25B documented savings though, it's more like $71 / person.

But perhaps we need to question any numbers DOGE puts out.

There are already numerous calculation errors and it seems that DOGE doesn’t even know how to count and don't understand the contracts. From @electricfuture:

The next 3 biggest ticket items are all USAID contracts listed as $655M each, so $2B total. Wow, pretty big. Wait, these are IDVs, not contracts. $655M is the entire set-aside, being triple counted. In the first 5 years, only $73M was awarded, and only 2 years remain. So we're down to $6.5B in savings, and an alarming trend emerges: @DOGE does not seem to understand how the government contracts they are canceling work. The savings they are claiming are not annual, but rather hypothetical if we spent every unobligated penny. And more importantly, they are just getting it wrong, with alarming consistency. These numbers are erroneous. This "select group of geniuses" has not double checked even the LARGEST items accounting for the bulk of their claimed savings. This is a sad, pathetic farce.

Here's the next biggest item: an IT services contract for the Social Security Administration worth $1B. That's a lot of savings! Well, again, this contract spanned 6 years. 80% has already been spent. Ah well, more like $240M in savings spread over the next 3 years. $80M/year. In 2023, this contract funded 1000 FTEs working $100/hr. Did we need 1000 SWEs working on SSA infrastructure? Probably not - these could be valid savings (disclaimer: no idea what they actually did). But worth noting that these cuts will impact many private sector jobs as well. And if anyone is curious, there are currently 17 lines that say "SEE FPDS" rather than the savings amount, I guess because their automated scraping failed. I did it manually and it took roughly ~10 minutes. But that's too much to ask of super geniuses working 120 hours/week!

And it's worse than it seems. DOGE has a blind focus on cost over value which means Elon has recently set his sights on deleting the CFPB. But it saved the country $17B in unnecessary overdraft fees, and we've saved ~$800M from deleting them - terrible ROI.

Maybe now is the right time to remind everyone that the only "savings" that seems to actually be true is the ~10% job cuts in the Federal government (though it’s not even clear the US federal government is that inefficient). And while layoffs sometimes a necessary thing, as any tech company founder will often claim, the method by which they've approached them feels extremely haphazard and not data-driven at all. Some examples I've seen:

• Firing employees simply because they've been at the company < 2 years rather than pushing for early retirement for highly tenured employees that likely earn more and can contribute less

• Firing employees simply because they are considered "probationary" even though the term also applies to those that were recently promoted

• Mass firings that seem to be related to oversight of companies that key government officials own or manage

With ~10% layoffs, we'll end up saving over $50B a year, which is definitely not nothing, but only accounts for ~$100 / person / year in cost savings. Meanwhile, this move energizes not just a directly affected 200K block of former federal workers but also a newly fearful 2.2M (who have had to fear for their job for the first time likely ever) into potentially voting for change. Is it worth it?