There are about 10,000 people in tech who have already hit retirement-level wealth — $20 million or more — thanks to the AI gold rush. That’s not a projection from a hedge fund or a speculative investor. It’s a back-of-the-envelope calculation from Deedy Das, a partner at Menlo Ventures, and it’s the most concrete number we’ve seen to date describing the AI wealth divide. As of May 17, 2026, that split isn’t just theoretical. It’s reshaping careers, companies, and the mood in San Francisco.
Key Takeaways
- About 10,000 people — founders and employees at OpenAI, Anthropic, Nvidia — have accumulated $20M+ in wealth.
- Meanwhile, software engineers in high-paying jobs — even at $500K annually — realize they’ll never reach that tier without equity.
- Layoffs are accelerating across tech, deepening uncertainty about the future of software careers.
- Many engineers feel their skills are obsolete, creating a deep malaise about work and relevance.
- Deva Hazarika pushed back, calling the complaints of tech workers “tone-deaf” given their privilege.
The AI wealth divide isn’t a bubble — it’s a split
You don’t need to believe in AI hype to see the damage it’s done to morale. The boom isn’t lifting all boats; it’s building yachts for some and leaving others to wonder if their life’s work still matters. That’s the core of what Das described in his original report. He’s not forecasting collapse. He’s observing a fracture — already here — between those who got in early and those who didn’t.
It’s not that people are losing money. It’s that they’re realizing they won’t ever gain what others already have. And that’s a different kind of pain. The $20 million threshold isn’t arbitrary. It’s the line where you stop needing to work. When you cross it, you’re free. When you’re stuck below it — even at $300K a year — you’re just another cog. That’s the new reality.
And it’s concentrated. OpenAI, Anthropic, Nvidia — these names keep coming up. You don’t have to be a founder. You could’ve been employee #50 at one of these shops and still walked away with generational wealth. But if you’re at a mid-tier SaaS company or a legacy tech firm, you’re watching the train leave without you.
The first wave of AI wealth started forming as early as 2020, when OpenAI pivoted from a non-profit to a capped-profit model and brought on Microsoft as a strategic investor. That deal, worth $1 billion initially, set the template: access to massive compute, institutional backing, and tight ownership of foundational models. Employees who joined before the GPT-3 release in mid-2020 — even in engineering, safety, or operations roles — ended up with equity packages that exploded in value once the commercial applications took off.
Anthropic followed a similar arc. Founded in 2021 by former OpenAI researchers, it raised $580 million in its first year and secured Google as a major investor by 2023. Early hires there, many of whom came from top AI labs or elite PhD programs, were granted stock options priced at fractions of a cent. By 2025, those same shares were trading in secondary markets at valuations that pushed individual stakes well into the tens of millions for some.
Nvidia’s rise was different — more industrial, less mysterious. But just as significant. The company had spent over a decade refining its GPU architecture for gaming and scientific computing. Then, in 2023, demand for H100 chips exploded. AI labs needed them to train large models, and there was no alternative. Nvidia’s revenue jumped from $26.9 billion in fiscal 2022 to $87.4 billion in 2025. Its stock price followed, lifting long-term employees and executives into untouchable financial territory. Engineers who joined in the early 2010s, product managers who stayed through quiet years, even factory supervisors in Taiwan — all saw their RSUs multiply beyond anything they’d imagined.
San Francisco is frenetic — but not in a good way
Das called San Francisco “pretty frenetic right now.” That’s an understatement — and a telling one. Frenetic isn’t excitement. It’s anxiety. It’s founders rushing to launch, engineers refreshing layoffs.fyi, VCs chasing term sheets like they’re expiring. The city isn’t buzzing with optimism. It’s vibrating with desperation.
Coffee shops in SoMa and Dogpatch used to hum with startup pitches. Now they echo with resignation. People still meet, still network, but the tone has shifted. It’s not “What’s your idea?” It’s “Do you know anyone hiring?” Or worse: “Did you get in anywhere early?”
The housing market reflects the split. Median rents in SF dipped slightly in 2025 after years of growth, but luxury units — those north of $10,000/month — remained fully occupied. The people cashing out aren’t leaving. They’re upgrading. Meanwhile, younger engineers and laid-off contractors are doubling up, moving to Oakland, or leaving the Bay entirely. The city feels thinner, more polarized.
Layoffs aren’t slowing — they’re accelerating
May 17, 2026, isn’t a quiet moment in tech. It’s another week with multiple layoffs across mid-sized AI startups and established engineering orgs. Companies that raised big in 2024 are now cutting staff because revenue hasn’t caught up to valuation. The cost of running large models keeps rising. Profitability feels further away than ever.
One startup that raised $200 million in 2024 on a $1.2 billion valuation just slashed 35% of its workforce. Their product? An AI-powered customer support agent. The problem? It didn’t reduce headcount enough to justify its own cost. Another company, building vertical-specific LLMs for legal firms, cut 50% after enterprise contracts failed to materialize. Sales cycles were too long. Budgets were tight. The ROI wasn’t obvious.
And it’s not just junior devs getting hit. Senior engineers — people with 15 years of experience — are finding themselves on the wrong side of the AI shift. Their resumes don’t scream “LLM ops” or “fine-tuning at scale.” They built APIs, microservices, databases. Solid work. But now it feels like legacy code in a world that wants prompt engineers and alignment specialists.
Skills that were valuable yesterday are optional today
Here’s what keeps engineers up at night: the fear that their life’s work doesn’t matter anymore. You spent 10 years mastering distributed systems. Now the new crop of grads are being hired to run RLHF pipelines. You wrote rock-solid backend logic. They’re designing agent workflows with autonomous loops.
It’s not that your skills are useless. It’s that they’re not valuable. Not in the way they used to be. And that distinction eats at people. Especially when they see peers — sometimes younger, less experienced peers — raking in equity from early AI bets.
- 10,000 people have $20M+ from AI gains
- Median software salary: ~$180K in SF (not enough to reach $20M without equity)
- Layoffs in May 2026: over 8,500 tech workers cut, per layoffs.fyi
- Nvidia’s market cap: $3.2 trillion — more than all but four countries’ GDP
- OpenAI and Anthropic have raised over $11B combined
Deva Hazarika’s rebuttal: you’re already lucky
Not everyone bought Das’s argument. Entrepreneur Deva Hazarika pushed back hard on X, saying most of the people in this conversation are “incredibly fortunate and can simply make a choice to be happy.” And he’s not wrong — not technically.
Compared to 99% of the world, a $300K job in tech is luxury. But that’s not how humans experience inequality. We don’t compare ourselves to villagers in rural India. We compare ourselves to the person in the next Slack channel who just exercised their options.
Hazarika’s take misses the point. It’s not about gratitude. It’s about fairness. It’s about watching a narrow band of people — often connected through the same schools, networks, and funding pipelines — capture almost all the upside. That’s not a meritocracy. That’s a network effect with stock options.
What This Means For You
If you’re a developer, you can’t ignore this. Your career path isn’t just about skill anymore — it’s about alignment. Are you building things that scale with AI, or compete against it? Are you close to equity, or just a salaried contributor? The people with $20M+ didn’t get there by writing clean code. They got there by being early and being inside the tent.
Consider a backend engineer at a well-funded fintech startup in 2023. She earned $350K total comp, worked hard, delivered consistently. But her equity was priced at a $2B valuation. The company hasn’t grown much since. Her stake is worth maybe $1.5M pre-tax — solid, but not life-changing. Meanwhile, a peer who joined Anthropic in 2022 at a lower base salary but with options priced at $4B now has a paper gain over $8M. The difference? Timing and access.
Or take a product manager at a legacy enterprise software firm. Her company adopted AI features slowly, outsourcing model work to third parties. When her division launched an “AI assistant” in 2025, it was built on GPT-4.1 and added minimal value. The stock didn’t move. Her bonus was cut. She started looking around — only to find that new roles demanded hands-on experience with vector databases, retrieval-augmented generation, and model distillation. Skills she hadn’t needed before. Skills she now has to learn on nights and weekends.
For founders, the gap is even starker. Imagine launching a dev tool in 2024 that helps teams manage API keys. Good problem, solid market. But in 2026, investors aren’t interested unless you’re building at the model layer or solving inference bottlenecks. Your Series A gets passed on — not because you’re bad, but because the money is flowing elsewhere. The wealth engine has shifted. You’re building apps on top of a platform, and platforms keep all the value.
And if you’re feeling that “deep malaise” Das described — that confusion about your future — you’re not alone. But the answer isn’t to quit. It’s to pivot. Learn how AI changes your stack. Get close to the data, the models, the deployment. Not because you’ll become a billionaire, but because you’ll stay relevant.
What Happens Next
The next 18 months will test whether this divide widens or starts to ease. Open-source models are improving fast. Llama 4, released in April 2026, runs efficiently on consumer hardware. Small teams can now fine-tune powerful models without relying on OpenAI or Anthropic. That could democratize access — at least technically.
But access isn’t the bottleneck anymore. It’s distribution, trust, and scale. Who do enterprises trust with their data? Who can guarantee uptime for critical workflows? Who has the sales force to close nine-figure deals? That advantage still belongs to the incumbents.
We’re also seeing a quiet migration. Engineers who missed the early AI wave are joining smaller labs, open-source projects, or starting their own tiny AI shops. Some are experimenting with agent-based startups, automated workflows, or niche vertical models. It’s possible one of them breaks through. But the capital, talent, and mindshare still orbit the core players.
Another question: what happens when the first wave of AI billionaires starts angel investing? They’ll fund people they know — former colleagues, friends, schoolmates. That could reinforce the existing network, or — if done intentionally — open new doors.
Will the next wave of AI wealth be broader — or will it deepen the divide? The tools are more accessible now, but the rewards still flow to the few who control the platforms.
Sources: TechCrunch, The Information
