Post

Premium Newsletter: The Hater's Guide To NVIDIA Part 2, the story of how it became the AI Bubble's GE Capital, propping up unsustainable companies and conning hyperscalers into buying GPUs to restart growth in the post-ZIRP economy. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

wheresyoured.atPremium: The Hater's Guide To NVIDIA (Part 2)For a little under a year, everyone — myself included — has compared NVIDIA to Enron, largely because NVIDIA insisted, in detail, that it was nothing like Enron, WorldCom, or Lucent, a potent example ...

With concerns about NVIDIA’s circular financing growing, we’ve seen the company insist it’s nothing like Enron, Worldcom, or Lucent. Perhaps not, but it’s still engaged in some weird indirect vendor financing shenanigans. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

NVIDIA has become far more of an asset manager or a bank than a technology company, with AI and GPUs existing only as tools to manufacture demand from customers who increasingly can't afford to pay for them, if they ever could. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

This is all very reminiscent of GECAS — GE’s airplane leasing wing, which, under Jack Welch, became the largest purchaser and lessor of airplanes for thousands of airlines that couldn't afford them, often backstopping their companies. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

GECAS took money from investors, which it then used to buy planes (often with GE engines), which it would then lease to airlines that couldn’t afford them — many of which couldn't survive without its bailouts. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

Obviously, this wasn’t sustainable, and in 2008 things came to a head. The problem was that GECAS was artificially holding up the aerospace industry by manufacturing demand. It couldn’t do this forever. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

What separates GECAS from NVIDIA is that nobody questions the need for planes, and the aerospace and remaining airline industry adjusted after several rocky years. Demand for AI services is pathetic outside of unsustainable AI labs. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

NVIDIA is smart, and worked out a way to do vendor financing without being the bank, using the FOMO around AI to con financial institutions into funneling billions into neoclouds by guaranteeing them revenue and funding. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

NVIDIA's greatest con is that it operates as the vendor, customer and financier-by-proxy for every neocloud, all without having to hurt its credit. Financial institutions are ignorant, and see circular financing as a good thing. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

NVIDIA doesn’t have to prove that AI is an industry with distinct demand that will exist long-term, it just has to show that it has some long term customers right now. Ratings agencies have failed to protect investors from neoclouds. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

The AI bubble truly started when every major hyperscaler's growth slowed in 2022, including NVIDIA. Once ChatGPT went viral, Jensen Huang would point to Microsoft and say "this is the iPhone moment for AI," giving everyone AI mania. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

People forget that NVIDIA had hit a wall in FY2023, and ChatGPT was its savior too. Every hyperscaler's growth had slowed that year, and suddenly the markets started rewarding them for buying all these GPUs. It was the perfect storm. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

Let's be blunt: hyperscalers did not restart growth by buying GPUs or "doing AI." What actually happened was the more GPUs they bought, the higher their stocks ran. Revenue growth has either stalled or declined over the last decade. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

70% of hyperscaler AI revenues are from two unsustainable companies, and demand does not exist at the scale that NVIDIA is selling GPUs. The world has been tricked by the idea that hyperscalers wouldn't make such a catastrophic mistake. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

7/8/26, 6:02 pm7 reposts51 likes

For NVIDIA to meet analyst expectations, it must sell $393 billion (mostly in GPUs) in FY2027 and $565.7 billion in FY2028. 50%+ of that comes from hyperscalers. Its future is based on the whims of executives getting few returns. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

People don't fully understand the hole that NVIDIA has dug for itself. While it’s unlikely to die, its growth projections depend on hyperscalers, flimsy Neoclouds, and revenues from unsustainable AI labs that must grow in perpetuity. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

Jensen Huang sold hyperscalers on the promise that buying GPUs would create infinite growth, and bullied them into buying more when that growth didn’t appear. At some point, something breaks. The more you buy, the more you lose. www.wheresyoured.at/premium-the-haters-guide-to-nvidia-part-2/

@bonfire25.bsky.social

Isn't a big part of the problem is EXPENSE of these data centers and GPUs that will inevitably become OBSOLETE long before the hyperscalers long term debt is paid?? Every time Jensen introduces another NEW GPU module for a lot more money, I ask myself how can this biz model be sustainable?

twistmeyer.bsky.social

@twistmeyer.bsky.social

I can't help but wonder if the retrenchment of companies like Sony and Microsoft on their games businesses isn't influenced by GPUs being prohibitively expensive and rare to gamers.