Who’s Actually Getting Rich?
Billions are being shoveled—sorry, invested—into AI startups like OpenAI and Anthropic. But who’s actually making money? Spoiler: It’s not the companies building robot brains. It’s the ones renting out the “brains” by the hour and selling the silicon shovels.
Just like in the original California Gold Rush, the dreamers with shovels are broke. The merchants selling those shovels? They’re buying yachts shaped like shovels.
This article digs past the AI hype to answer one burning question: Who’s really profiting from the generative AI boom—and who’s just setting investor cash on fire for GPU time?
Funding Reality: OpenAI vs Anthropic
Let’s look at the financial reality of two leading generative AI startups, OpenAI and Anthropic. Despite the hype, both companies are burning cash at an extraordinary rate, relying on deep-pocketed backers.
The table below sums up their funding, revenue, and expenses since 2022:
| Metric | OpenAI | Anthropic |
|---|---|---|
| Total funding (since 2022) | ~$10.3 B (via Microsoft & venture capital) | ~$12.4 B (via Amazon, Google, VC) |
| AI-related revenue (2023–2024) | ~$1.2 B (estimated total) | ~$1 B (estimated annual run-rate by end of 2024) |
| Capex (infrastructure via partners) | Co-building Azure supercomputers with Microsoft | Runs on AWS/Google Cloud (rents capacity) |
| Business model | SaaS (ChatGPT & API), licensing, integrations (e.g. MS Copilot) | SaaS (Claude API, Claude via AWS/Google), cloud partnerships |
| Cash burn rate (est.) | ~$500 M–$1 B per year (R&D, infra, talent) | ~$1.5 B+ per year (massive model training & scale-up) |
In plain English: OpenAI and Anthropic have raised astronomical sums to chase AGI (Artificial General Intelligence) fever dreams. Yet, their revenues so far are comparatively minuscule. Both are essentially subsidized by “Big Tech sugar daddies” (Microsoft for OpenAI; Amazon and Google for Anthropic), who provide cash and cloud credits to keep the AI research going. Each startup’s cash burn far exceeds its income, meaning they are a long way from self-sustainability.
Where’s the Gold? (Hint: Not in Startup Bank Accounts)
OpenAI will reportedly make about $4 B in revenues in 2025. Their computing costs? Around $5 B. That’s right—every time you ask ChatGPT to write a limerick, Sam Altman’s wallet cries a little.
Over 90% of OpenAI’s 500 million users pay nothing. Anthropic’s Claude AI? Growing fast, but not fast enough to justify burning $1+ billion a year. Enterprises are dabbling in genAI, but most are still on the fence, worried about accuracy, cost, and whether their data will end up training the next chatbot that calls them “bro.”
The bottom line: OpenAI and Anthropic need a stampede of paying customers, not just buzz. Until then, they’re stuck with “strategic investors” who are basically paying their cloud bills and hoping for a miracle.
The Shovel Sellers: Cloud & Chip Giants Cashing In
While AI startups are busy tossing investor cash into the bonfire of “innovation,” there’s a group quietly counting their gold coins in the back room: the infrastructure giants. Yes, the real jackpot winners of the AI gold rush aren’t the ones building clever chatbots that can write Shakespearean sonnets about your cat. They’re the ones renting out the digital pickaxes and selling the computational shovels—at prices that would make a Silicon Valley landlord blush.
Let’s break it down, gold rush style: in the 1800s, it wasn’t the miners who got rich. It was the merchants selling picks, shovels, and the world’s most durable denim. Fast forward to today, and the new Levi Strausses are Amazon (AWS), Microsoft (Azure), Google Cloud, and NVIDIA. Their business model? Sell the tools, watch the prospectors sweat, and never, ever run out of snacks in the executive lounge.
Amazon Web Services (AWS): The Cloud Kingpin
- Revenue Rocket: AWS hauled in $90.8 B in 2023, then cranked it up to $107.6 B in 2024. If AWS were a country, it would have its own Olympic team—probably all robots.
- Spending Spree: Amazon’s CFO announced over $100 B in planned 2025 capex, mostly for AWS infrastructure. That’s more than OpenAI has raised in its entire existence, just to keep the server farms humming and the GPU vending machines stocked.
- AI Demand: Every AI startup, from the next ChatGPT wannabe to your neighbor’s “AI-powered” toaster, is clamoring for AWS compute. The result? Amazon’s cloud is basically a money-printing fog machine, billowing out invoices faster than you can say “serverless.”
Microsoft Azure: The Supercomputer Superstore
- Intelligent Cloud: Microsoft’s cloud division (mostly Azure) raked in $96.8 B in 2023, with growth rates that would make even ChatGPT blush—over 30% year-on-year.
- AI Arms Race: Microsoft isn’t just building clouds; it’s building AI superclusters for OpenAI and anyone else with a sufficiently large credit card. Projected capex for 2025? Around $80 B, a number so large it could fund a small moon colony—or at least a really fancy break room for Azure engineers.
- Already Winning: While OpenAI dreams of $13 B in revenue, Microsoft is already there, selling cloud time to every AI hopeful and their dog. (Yes, even the AI-powered dogs.)
Google Cloud: Profitable at Last, and Loving It
- Revenue Leap: As the third-biggest provider, Google Cloud pulled in $33.7 B in 2023 and is expected to hit $44 B in 2024. After years of “investing in growth” (translation: losing money), it finally turned a profit—just in time for the AI boom.
- AI Everything: Google is pouring billions into AI hardware (like TPUs) and cloud platforms (Vertex AI, anyone?). Alphabet’s 2025 capex? Roughly $75 B, mostly earmarked for—you guessed it—more data centers.
- Market Surge: The global cloud market hit $330 B in 2024, with analysts blaming at least half that growth on generative AI. Anthropic might pay Google for compute, but Google is the one laughing all the way to the (cloud) bank.
NVIDIA: The Pickaxe Kingpin
- GPU Goldmine: If AWS and Azure are selling shovels, NVIDIA is hawking diamond-encrusted pickaxes. Its A100 and H100 chips are the must-have bling for every AI startup, Fortune 500, and probably your local library’s new “AI-powered” book sorter.
- Financial Frenzy: In one 2023 quarter, NVIDIA’s data-center revenue soared 171% year-over-year to $10.3 B. By late 2024, it briefly became the world’s second most valuable company—right behind Apple and just ahead of “Your Startup Here.”
- Growth on Steroids: Five straight quarters of triple-digit revenue growth. Q4 2024 revenue? Projected to surge 72% year-on-year to $38 B. Investors see NVIDIA as the surest bet in the AI casino, because every new chatbot, image generator, and “AI-powered” coffee maker needs a mountain of NVIDIA chips.
- Market Cap Madness: By early 2025, NVIDIA’s valuation was second only to Apple, leaving Google and Meta eating its silicon dust. When it comes to selling shovels, NVIDIA doesn’t just sell the tools—it sells the gold-plated, WiFi-enabled, limited-edition tools.
Other “Arms Dealers” in the AI Arms Race
- Oracle & Friends: Not to be left out, Oracle and other cloud hosts are jumping into GPU-as-a-service, boasting impressive growth as everyone and their grandmother wants a slice of AI pie.
- AMD: Eager to dethrone NVIDIA, AMD is rolling out its own high-powered GPUs, hoping to snag some of the gold dust flying around the server rooms.
- Data Center Ecosystem: Even the companies selling electricity, cooling systems, and backup generators are cashing in. If it keeps the server racks from melting, someone’s making money off it.
- No Model, No Problem: The beauty of being an infrastructure vendor? You get paid whether or not the latest AI model is a world-changing genius or a glorified autocorrect that thinks “annual budget” means “animal buffet.” As long as the AI arms race continues, the meter keeps running—and the cash keeps flowing.
Conclusion: The Irony of the AI Gold Rush
It’s a delicious irony: while AI startups evangelize a future of intelligent machines and scramble to monetize mind-blowing tech breakthroughs, the immediate financial winners are relatively “boring” B2B providers – cloud server landlords and chip peddlers. OpenAI and Anthropic might eventually strike gold with a truly killer app or dominant platform (that’s certainly what their investors are betting on). But right now, it’s the pick-and-shovel sellers who are laughing all the way to the bank.
The situation brings to mind the old quip about the Gold Rush: “You can mine for gold, but you’ll probably just end up washing dirt – the guy selling you the shovel is the one who makes a profit.”
Today’s cloud and semiconductor giants are those shovel sellers. They don’t particularly care which AI startup wins the model accuracy contest or lands the biggest enterprise deal – they profit as long as someone is training gargantuan models or spinning up AI workloads on their platforms.
So, as generative AI fever continues, keep an eye on who’s actually making the money. The next time you hear about a fancy new AI model that writes poetry or code, remember: somewhere, an AWS data center is humming, accruing charges to the startup using it. And somewhere, NVIDIA’s CEO is selling another truckload of GPUs to feed that model. In this gold rush, the smartest strategy might be to sell shovels, not chase nuggets of AI gold. The cloud providers and chipmakers have figured that out – and their bottom lines are gleaming because of it.
Cutting Through the AI Hype with Humor and Insight
If you want to dive deeper into the baffling and downright stupendous world of AI—beyond the gold rush hype—check out my book, Artificial Stupelligence. It’s your backstage pass to the AI circus, packed with laughs, eye-opening insights, and a healthy dose of skepticism. Because in the wild west of artificial intelligence, sometimes the smartest move is knowing when to sell the shovel instead of chasing the fool’s gold.
Sources:
- The Economist – reporting on AI project abandonment and enterprise disillusionment
- McKinsey & Company – AI adoption statistics, enterprise readiness, and productivity impact
- Synergy Research Group – global cloud infrastructure market data
- Microsoft, Amazon (AWS), and Google (Alphabet) – cloud revenue and capex figures
- NVIDIA – AI GPU sales and data center growth
- WIRED & Bloomberg – coverage of AI talent compensation wars and hiring trends
- Reuters & Financial Times – funding and revenue details for OpenAI and Anthropic
