Oracle trading like a penny stock

And the AI-bubble

Oracle, a $922 billion company, decided to cosplay as a penny stock this week, delivering a spectacular 36% single-day surge that would make the most volatile cryptocurrency blush with envy. The company managed this impressive feat despite the minor inconvenience of missing both revenue and earnings estimates—because who needs actual performance when you have AI promises? The $244 billion market value increase in one trading session was so absurd it catapulted Larry Ellison past Elon Musk to become the world’s richest person, proving that in 2025, the fastest way to extreme wealth is promising impossible things about artificial intelligence.

The magic catalyst behind this financial wizardry? Oracle announced $455 billion in total new cloud contracts, headlined by a $300 billion deal with OpenAI—a company that burns through money faster than a teenager with their first credit card. This arrangement requires OpenAI to pay Oracle $60 billion annually starting in 2027, which would be merely ambitious if OpenAI weren’t currently hemorrhaging $2.2 billion monthly while begging investors for an additional $40 billion just to keep the lights on. In just 30 months since ChatGPT’s launch, OpenAI has managed to transform $66 billion in venture capital into a monthly cash furnace that would make even the most profligate government agencies envious. But hey, why let basic arithmetic spoil a perfectly good bubble?

Welcome to the upside-down economics of artificial intelligence, where every transaction is designed to destroy value at the user level while concentrating wealth at the hardware level. When you innocently pay $1 for an AI application, you’ve unwittingly triggered a reverse money machine that would make perpetual motion inventors weep with joy. That dollar magically multiplies as it travels upward: application companies pay $5 to model providers like OpenAI, which remits $7 to cloud infrastructure providers like Oracle, who ultimately hand over $13 to GPU manufacturers like Nvidia. It’s essentially a pyramid scheme, except the pyramid is upside down and everyone pretends this makes perfect sense.

Of course, Oracle knows a thing or two about creative accounting during tech bubbles. Back in 1990, Larry Ellison’s company discovered the delightful accounting technique of booking future sales to meet current quarterly targets, inflating earnings by a modest 30%. This innovative approach to mathematics eventually forced Oracle to restate financials twice while laying off 10% of its workforce—a small price to pay for temporary quarterly glory. Now, three decades later, Oracle projects cloud revenue growing from $18 billion to $144 billion by 2030 while simultaneously eliminating over 10,000 jobs. It’s almost as if there’s a playbook for this sort of thing, and Larry Ellison wrote the first edition.

The beneficiary of this magnificent reverse value chain sits comfortably at the top: Nvidia, where Jensen Huang has transformed from graphics card salesman to the high priest of GPU economics. His company captures the lion’s share of every AI transaction, which is impressive efficiency if you ignore the part where the entire system requires exponentially more investment than it generates in user value. The Magnificent Seven technology companies plan to spend $364 billion on data centers in 2025 alone, creating the largest coordinated capital expenditure in technology history funded by user demand that generates a fraction of the required revenue. It’s like building a highway system for cars that don’t exist, except the cars also cost more than houses and depreciate faster than fresh produce.

The real beauty of this system becomes apparent when examining actual AI implementation results. MIT research reveals that 95% of corporate AI projects deliver zero return on investment despite $30-40 billion in enterprise spending. This means companies are enthusiastically purchasing expensive GPU infrastructure for applications that measurably fail to improve productivity or profitability. It’s the equivalent of buying lottery tickets with a 95% guaranteed loss rate, except somehow this makes the lottery ticket manufacturers incredibly wealthy and everyone agrees this represents cutting-edge business strategy.

The market concentration behind this magnificent pyramid of value destruction has reached truly spectacular levels. Just 38 AI companies now control 49.4% of the S&P 500’s total market capitalization, which is perfectly normal behavior for a healthy, diversified economy. Oracle’s transformation from boring enterprise software company to AI infrastructure speculator represents the kind of bet-the-company pivot that either creates legends or cautionary tales. The company’s P/E ratio of 54.15 sits at a comfortable double its historical average, because why settle for reasonable valuations when you can have impossible ones?

Venture capital flows reveal the true genius of this system, with AI startups capturing $100 billion globally in 2024 despite documented implementation failures across industries. Over 370 AI “unicorn” companies are valued above $1 trillion collectively, concentrating more speculative capital in fewer entities than during the quaint, distributed dot-com boom. This time it’s different, naturally, because instead of spreading the delusion across thousands of startups, we’ve efficiently concentrated it in fewer companies making larger bets on technology that demonstrably doesn’t work for most use cases.

The sustainability question becomes particularly amusing when examining Oracle’s infrastructure requirements. The company’s $92.6 billion debt load—a debt-to-equity ratio of 427% compared to Microsoft’s quaint 32%—funds data centers that face rapid technological obsolescence. GPU technology cycles of ~3 years mean these massive capital investments depreciate faster than smartphones, while unproven demand sustainability makes long-term planning essentially a sophisticated form of gambling. Oracle’s executives seem keenly aware of these risks, with CEO Safra Catz selling ~$400 million in shares during June’s AI enthusiasm, because nothing says confidence in your AI strategy like immediately cashing out.

Oracle’s penny stock performance at trillion-dollar scale represents the perfect microcosm of our current AI bubble. Larry Ellison’s elevation to world’s richest person through a single day’s speculation on contracts with bankrupt customers demonstrates how bubble dynamics concentrate extreme wealth while the underlying business models remain fundamentally broken. The AI money machine that runs in reverse has created the largest misallocation of capital in technology history, with Oracle serving as both willing participant and perfect cautionary tale about what happens when mathematical impossibility meets institutional amnesia about previous bubble dynamics.

The ultimate irony is that artificial intelligence, supposedly the pinnacle of logical reasoning, has created an economic system that would be rejected by a middle school math class. This inverted pyramid of value destruction can only sustain itself as long as participants maintain religious faith that the next breakthrough will somehow reverse the fundamental laws of economics—a belief system that makes previous technology bubbles look like models of rational skepticism.