AI & the Artificial Revenue Illusion

OpenAI can’t afford to pay cash upfront for the massive compute it needs, so it strikes a deal with AMD. AMD issues warrants, the right for OpenAI to buy AMD shares later at preset prices, tied to future deployment milestones. The deal is announced, AMD’s stock jumps 25%, and suddenly both companies look stronger on paper. OpenAI hasn’t spent real cash yet, but it’s secured future compute capacity and holds equity that could rise in value as AMD grows from the partnership.

It’s a clever form of vendor financing, a loop where financial incentives, stock appreciation, and strategic supply all reinforce each other. Wall Street calls this kind of structure circular trading or round-tripping: capital moves in a circle, creating value through alignment and timing rather than direct payment. Everyone wins, until the loop stops spinning.

Both benefit from a reinforcing cycle of growth, valuation, and supply.

• OpenAI commits to buy billions in AMD chips over time.

• AMD grants OpenAI warrants to buy AMD stock at fixed prices if deployment targets are met.

• The market reacts to the partnership; AMD’s stock surges 25%.

• The higher valuation gives AMD more leverage and capital to produce chips.

• OpenAI gains compute access without large upfront cash outflow.

Dangers of Circular Financing

Here's why the circular financing structure is fundamentally dangerous:

1. The Artificial Revenue Illusion: When Company A invests in Company B, and Company B uses that money to buy from Company A, who is actually the customer? The revenue appears real on financial statements, Nvidia can legitimately book billions in chip sales. But the economic substance is questionable at best. It's the financial equivalent of writing yourself a check and calling it income. Technically accurate, economically meaningless.

2. The Valuation Feedback Loop: The circular structure creates a self-reinforcing cycle that completely divorces stock prices from underlying economics:

• High valuations enable large investments
• Large investments enable purchases
• Purchases drive revenue growth
• Revenue growth justifies high valuations
• Higher valuations enable even larger investments

This works beautifully in one direction, upward. But it has an equally powerful dynamic in reverse. If any link in the chain weakens, the entire structure can collapse simultaneously. This isn't theoretical; it's exactly what happened to WorldCom and the entire telecom sector during the .com boom & bust in the late 90's.

3. The Liquidity Mirage: These arrangements create the appearance of robust demand and healthy cash conversion, but it's illusory. Adrian Cox at Deutsche Bank specifically flags his concern about "cross ownership structures" emerging in the AI sector, precisely the kind of interconnected dependencies that transform localized problems into systemic crises.

Real demand means customers with independent funding sources willingly pay for your product because it creates value greater than its cost. Circular funding means you're essentially buying from yourself with extra steps and financial engineering.

4. Distributed Risk Becomes Systemic Risk: When companies are interconnected through circular financing, a problem at any point in the chain propagates everywhere. In the AMD/OpenAI case, consider what happens if OpenAI stumbles and cannot make its chip purchase commitments → AMD's revenue projections collapse → AMD stock falls → OpenAI can't exercise profitable warrants → OpenAI can't fund operations → OpenAI's valuation craters → AMD's investment loses value → AMD's balance sheet takes a hit → AMD becomes more cautious about future investments → other AI companies can't secure funding

One crack becomes a cascade. This interconnectedness is precisely what made the 2008 financial crisis so devastating. Banks were so interconnected through derivatives and mortgage-backed securities that Lehman Brothers' failure nearly toppled the entire global financial system.

For the past two years, the AI investment frenzy has followed a predictable script. Whenever concerns about unsustainable valuations surface, they're immediately dismissed as pessimism or "not understanding the technology." Critics are told they're missing the forest for the trees, that AI is fundamentally different from past technological manias.

But something fascinating happened in August 2024. Adrian Cox at Deutsche Bank Research Institute noticed a spike in Google searches for "AI bubble", triggered by Sam Altman's admission that investors were "overexcited" and an MIT study showing 95% of AI pilots weren't working out. The concern was real, measurable, and growing. Then it vanished.

By September, searches for "AI bubble" had plummeted to just 20% of their peak. The fear evaporated faster than morning fog in Silicon Valley. Cox's analysis reveals something profound about market psychology: "The only thing we have to fear is a lack of fear itself. It is the lack of fear that can really lead to a bubble.". When everyone stops worrying, that's exactly when you should start.

The Historical Echoes Are Deafening

The late 1990s offer striking parallels. Telecom giants like Nortel, Lucent, and Cisco lent huge sums to carriers such as WorldCom, Global Crossing, Sprint, and AT&T, money used to buy their own equipment.

Mechanics: • Equipment maker lends $1B Carrier uses it to buy equipment → Lender records $1B in revenue → Stock surges → Cycle repeats. It worked, until borrowed money dried up and real demand never appeared. At some point the bubble bursted:

• WorldCom filed for bankruptcy ($104B in assets)
• Global Crossing folded
• Nortel fell from $124 to $0.67
• Lucent dropped from $84 to under $1
• Hundreds of billions evaporated

Then it was debt-based vendor financing on balance sheets. Today, it’s equity stakes and warrants, even murkier and harder to regulate.

Key question: Is real demand growing fast enough to justify the infrastructure, or are we building a beautiful edifice with no foundation?

In last week's blog we highlighted that the .ai situation if very different from the .com situation in the late 90's. Many argue circular financing reflects real, growing demand, not manipulation.

When Does the Machine Break?

The circular financing scheme becomes a crisis when any of these triggers occur:

Demand Reality Check: If businesses don't start generating sufficient AI revenue to justify infrastructure costs, the entire premise collapses. Right now, companies are spending on AI in anticipation of future returns. That patience isn't infinite. An MIT study found that 95% of companies using AI pilot programs aren't seeing productivity returns. If that doesn't improve dramatically, CFOs will stop approving AI expenditures. When that happens, the circular money machine has no fuel.

Market Sentiment Shift: You could have a shift in sentiment in the marketplace. For whatever reason, the stock market is giving folks higher valuations when they say they're going to spend more. If the market suddenly decides that spending announcements signal desperation rather than confidence, valuations crater. And unlike a normal market correction, the circular financing structure means one company's problem becomes everyone's problem.

Accounting Scrutiny: If regulators or auditors start questioning these arrangements as genuine arm's-length transactions, companies might be forced to restate revenues, revalue investments, or face trading restrictions. The SEC has historically taken a dim view of circular trading arrangements. The fact that today's structure uses equity rather than debt doesn't necessarily make it more palatable to regulators, it just makes it less tested in regulatory frameworks.

Capital Markets Freeze If any major player can't access additional capital, through equity raises, debt markets, or new investors, the circular flow stops. And unlike a supply chain disruption that affects one company, this affects everyone simultaneously.

The Uncomfortable Truth

The circular money machine powering AI isn't inherently evil or fraudulent. It might even be necessary, the traditional venture capital model may be insufficient for the capital requirements of AI infrastructure. Perhaps this creative financing is how revolutionary technologies get built in an era when even trillion-dollar companies can't fund development alone.

But let's be honest about what we're watching: We're in the experimental phase of an entirely new financing model, deployed at unprecedented scale, with interconnected risks we don't fully understand, during what even optimists admit is a bubble.

The circular machine keeps spinning. The valuations keep climbing. The announcements keep coming. And deep in the back of everyone's mind is a simple question: When the music stops, who's holding the bag? History suggests it won't be the people who designed the system.