Could AMD’s Newest Chip & Software Challenge Nvidia’s Dominance?

We Encourage You to Click & Let Pink Floyd Play while Reading….

We have presented to a large number of Financial Advisors and institutions of many stripes.  The questions around AI are many.  They range from “what inning are we in?” to “how can you think this is a bubble when the big tech stocks today are so much healthier than in the dot.com bubble?”

We have no idea what inning this is.  We refuse to even guess. But we are certainly in a bubble!

If you’re thinking we have sour grapes….our asset management firm’s core products have “tracking error budgets” and, driven by data, have done well given the speculative mayhem at play.  We’ve consistently made up for our light-weighting of the tech oligopolies trading at nosebleed valuations by finding “bank-shots” that offer upside leverage to the tech mania while still providing a significant margin of safety.

Get our insights direct to your inbox: SUBSCRIBE

Let us be clear: We are grateful to the speculators that race to the reasonably priced high-quality stocks we own and send them soaring on news announcements of dubious merit.  See live fire examples in our trade commentaries on Oracle, SMCI, and many others in our Trade Commentaries.

So today’s piece is structured in three parts. 

• First – a solid page documenting the feckless financial shell-games underway.
• Second: with AMD’s earnings set to release on Monday, we compare Advanced Micro Devices’ newfound tech prowess with Nvidia’s technology and surface some surprising conclusions – is AMD pulling ahead of Nvidia?
• Third: we confess to having a short position in AMD despite readily acknowledging that the disparity in technology with the current chip leader, Nvidia, has shrunk to such an extent the 10x disparity in valuation may close….to the detriment of AMD short sellers like us.

Stupid? Maybe. Honest? Absolutely.  Let’s dive in!

The AI boom is a mirage propped up by circular financing and debt-fueled hype. Nvidia, Microsoft, Oracle, OpenAI, AMD, and others are swapping equity, revenue, and promises—often in closed loops that obscure real demand. [1] Nvidia’s $100 billion “investment” in OpenAI is matched by OpenAI’s commitment to buy Nvidia chips, creating a feedback loop of artificial demand and self-reinforcing valuations. [2] Oracle is spending $60 billion a year on a cloud deal with OpenAI—an amount OpenAI doesn’t yet earn—for data centers that haven’t been built and for electricity the grid cannot supply. [3] To finance it, Oracle is taking on over $100 billion in new debt, pushing its debt-to-equity ratio to 500%. [4]

The second rot lies in the vast overbuild of infrastructure justified by faith in a yet-to-be created digital deity, not cash flow. Giants are racing to construct trillion-dollar data center networks while adoption—especially among enterprises—remains tepid. [5] MIT found that 95% of generative AI pilots have failed to generate any return on investment, despite companies collectively spending $30–40 billion across more than 300 deployments. [6] Meanwhile, a narrow set of AI companies—chiefly Nvidia, Microsoft, and their ecosystem—now account for 75% of the S&P 500’s total return since the October 2022 bear market, 80% of its earnings growth, and 90% of its capital expenditures. [7]

This extreme concentration of returns is not evidence of broad innovation. It’s a single speculative trade distorting an entire market. There’s an ethical corrosion behind the exuberance.  Nvidia, AMD, Oracle, and Microsofts’ intertwined deals tilt the field toward oligopoly, undermining the open-market dynamism that once fueled tech revolutions. This concentration of capital and compute reflects not just a speculative bubble—but a moral one. The circular AI money loop rewards collusion over competition, which may leave smaller innovators starved for oxygen while leaving us vulnerable to disruption from more nimble foreign startups. [8] At the same time, these companies are inflating short-term profits by depreciating AI hardware over 5–7 years, even though the true useful life is closer to 2–3 years. [9]

The infrastructure is brittle. AI data centers have pushed grid demand to crisis levels, triggering record investment in transmission and backup systems while providing little in the way of sustainable jobs or wage growth. [10] These facilities are capital-intensive and people-light, with many running almost entirely automated once built. [11] The real economic multiplier is minimal. Yet companies like AMD and CoreWeave are racing to keep up with hyperscaler capex by taking on their own risky debt loads—despite unclear demand and uncertain returns. [12],^[13] Meanwhile, early-stage AI startups with no revenue or products are being valued in the billions simply for adding “AI” to their pitch decks. [14]

This is not sustainable. It’s not a revolution. It’s a debt-fueled echo of the dot-com collapse.  The dream of frictionless automation has curdled into an expensive, low ethics experiment rife with data contamination whose end goals are often dictated by AI’s high priests, cloistered in the billions they have grifted from actual published work. [15]

Don’t believe us?  A wonderful article in the Wall Street Journal worked hard to explain the circular funding of these arrangements.  Here is one tiny excerpt (and by no means the worst example) of the conflicts at play which also helps us pivot into the topic at hand today: Advanced Micro Devices vs. Nvidia.

AMD, a rival to Nvidia, was so eager to land OpenAI as a customer that it issued warrants for OpenAI to buy as much as 10% of AMD at a penny a share.  AMD said it expects tens of billions of dollars of revenue, but its basically paying OpenAI to become a customer.

CoreWeave, an AI-cloud infrastructure company that rents out data-center capacity, illustrates the industry’s complex ties.  Nvidia owns about 5% of CoreWeave and sells chips to CoreWeave.  Nvidia also committed to purchase any unsold cloud-computing capacity from CoreWeave through 2032, effectively backstopping its customer.

Meanwhile, CoreWeave’s biggest customer is Microsoft, which is an investor in OpenAI, shares revenue with OpenAI, buys chips from Nvidia and has partnerships with AMD.  OpenAI also is a CoreWeave customer and shareholder, having made a $350ml equity investment in the company before its initial public offering.

Phew!  That’s out of the way.   Now let’s look at AMD.  Trading at 77x non-cash earnings, AMD generated $2.6bn in free cash flow over the trailing 12 months. [16] The company’s enterprise value tallies to a whopping (or maybe far too modest?) $427 billion dollars.  That is 164x free cash flow.

As we will show – the high-hopes embedded in that multiple suggest to us that AMD might just be a bargain compared to its $5 trillion peer Nvidia…while still leaving AMD a terrific short.  The seeming conflict stems from AMD’s impending tech advantage in both wafers and software (cheap vs. NVDA) that, even if it all works as analysts hope, might still disappoint investors.  164x is not a multiple for the faint of heart!

The KCR team is happy to debate what a healthy capital structure looks like for a firm or firms.  But we’ll go out on a limb and say we think most people would agree that the below is intuitively troubling.