Follow the Money in a Circle

Nvidia crossed $40 billion in equity investments in 2026. The biggest chunks: $30 billion into OpenAI, $3.2 billion into Corning, $2.1 billion into IREN.

Here’s the pattern that has people uncomfortable: Nvidia invests in a company. That company buys Nvidia GPUs. Nvidia books the revenue.

Money out, GPUs out, money back in. Rinse and repeat at scale.

The Bloomberg Investigation

Bloomberg’s investigative team started pulling the thread in April. What they found isn’t illegal. But it raises questions that Nvidia’s 35x forward earnings multiple probably can’t absorb.

The core issue: when your biggest customers are also your investment portfolio, where does organic demand end and manufactured demand begin?

Nvidia’s answer is that these companies would buy GPUs regardless. The investments are strategic, not demand generation. Maybe. But $40 billion in strategic investments that happen to flow back as GPU purchases deserves more scrutiny than a hand-wave.

The OpenAI Deal: $30 Billion of Questions

The OpenAI investment is the elephant. Thirty billion dollars — the largest single equity investment in AI history.

OpenAI is Nvidia’s biggest customer. It’s also now one of Nvidia’s biggest investments. The company that sets GPU demand is financially entangled with the company that supplies GPUs.

This creates incentives that should make investors nervous. OpenAI has less reason to diversify away from Nvidia hardware. Nvidia has every reason to keep OpenAI dependent on its chips. The investment isn’t just capital — it’s a lock-in mechanism dressed up as a bet on the future.

Michael Burry Sees the Pattern

Michael Burry — the “Big Short” guy — doubled his short positions on both Nvidia and Palantir in Q1 2026.

Burry doesn’t short on vibes. He shorts when he sees structural problems that the market is ignoring. His thesis, pieced together from 13F filings and his cryptic social posts: the AI revenue cycle contains more circularity than the market has priced in.

He’s been wrong on timing before. He was early on the housing crisis too. Being early and being wrong look identical right up until they don’t.

The Numbers That Don’t Add Up

Let’s do the math on what “organic” GPU demand looks like when you strip out Nvidia-funded buyers.

Nvidia’s data center revenue in 2025: $115 billion. Companies in Nvidia’s investment portfolio accounted for an estimated 30-35% of that revenue. That’s $35-40 billion in GPU sales to companies Nvidia is simultaneously funding.

Is all of that circular? No. These companies have other revenue sources and other investors. But the entanglement is deep enough that separating organic demand from investment-fueled demand is genuinely difficult.

That ambiguity is the problem.

Data Center Backlash Is Real

While Wall Street debates circularity, something more tangible is happening on the ground. Fourteen states now have active legislative efforts to restrict data center construction.

The complaints are consistent: massive power consumption, minimal local jobs, tax incentive abuse, noise pollution, and strain on water resources for cooling.

Virginia — home to the largest concentration of data centers on Earth — passed restrictions on new builds near residential areas. Ohio blocked a proposed 500MW facility. Local opposition groups in Texas, Georgia, and Arizona are organized and growing.

This matters for Nvidia because every GPU sold needs a data center to live in. Regulatory friction on data center construction is indirect demand friction on GPU sales.

PulteGroup and the Decentralization Bet

Here’s a wildcard nobody expected: PulteGroup, the homebuilder, is testing “home data center nodes.” Small, residential-scale compute units that distribute AI workloads across homes instead of concentrating them in massive facilities.

It’s early. The units are small. But the concept attacks the core vulnerability of the centralized data center model — the political and physical limits of concentrating thousands of megawatts in one location.

If distributed compute works at scale, it reshapes who buys AI chips and how. Nvidia wins either way on silicon, but the margin structure and customer concentration would look very different.

What Circular Financing Breaks

Circular financing doesn’t break when times are good. It breaks when growth slows.

If AI spending plateaus — even temporarily — the companies Nvidia invested in face revenue pressure. They cut GPU orders. Nvidia’s revenue drops. The value of Nvidia’s equity investments drops. Nvidia has less capital to invest. The virtuous cycle becomes a vicious one.

This is exactly what happened with SoftBank’s Vision Fund in 2022. Invest in companies, those companies spend on growth, growth metrics justify higher valuations, higher valuations justify more investment. Until it stopped.

The Market Doesn’t Care — Yet

Nvidia’s stock is near all-time highs. The market has decided that $40 billion in AI investments is bullish, not suspicious.

Markets can stay irrational longer than shorts can stay solvent. Burry knows this better than anyone.

But the combination of Bloomberg digging, Burry shorting, data center backlash building, and the sheer mathematical circularity of the investment-to-revenue pipeline — it adds up to a risk profile that the current valuation doesn’t reflect.

The question isn’t whether Nvidia makes great chips. It does. The question is whether $40 billion in investments in your own customers is demand creation or demand discovery.

That distinction is worth about a trillion dollars in market cap.