Michael Burry is leaning harder into his bearish case on Nvidia after the largest company on earth beat earnings yet again on Wednesday, something Cryptopolitan extensively reported.
In a Thursday Substack newsletter, Michael pointed to Nvidia’s purchase obligations jumping to $95.2 billion, from $16.1 billion a year earlier. He also cited total supply obligations, including inventory and purchase agreements, at about $117 billion, which he said sits close to the company’s annual operating cash flow.
Michael calls the supply lock-in a business plan problem
On Wednesday’s fiscal fourth-quarter earnings call, Chief Financial Officer Colette Kress said inventory rose 8% from the prior quarter.
Colette also said Nvidia had “strategically secured inventory and capacity to meet beyond the next several quarters, further out in time than usual.”
Michael took that as a company locking in supply before it can truly know how strong future demand will be. He wrote:-
“What is happening now is not temporary. It is no export shock. It is not even external. This is coming from within the business plan. This new reality reflects a deliberate decision to lock up supply chain capacity further than Nvidia has ever done before.”
Michael compared today’s setup to Cisco Systems during the dot-com boom in the late 1990s and early 2000s. He pointed to 2000 and 2001, when Cisco secured large supply commitments to support rapid growth expectations.
When corporate tech spending later fell, Cisco ended up stuck with extra inventory and supplier contracts it could not use. Cisco then wrote down billions of dollars, and the stock dropped hard.
Michael said: “This is not business as usual. This is risk. Back in 2000-2001, Cisco extended purchase commitments with its suppliers to ensure capacity for that 50% annual growth Cisco expected.”
He noted that Nvidia’s profit margins are above 70%, higher than what Cisco had in that period. Michael said that could soften the downside.
But he also wrote that those margins have been helped by unusually strong demand and the company’s ability to raise prices. He warned, “That type of margin would likely revert quickly with a shift in demand.”
Jensen says the networking unit is now a giant business
On the same earnings call, Chief Executive Officer Jensen Huang talked about the company’s networking business inside its data center unit. The segment is smaller than the computer side, but it has been getting more attention as AI buildouts get bigger.
Jensen said, “We’re … now the largest networking company in the world.”
He split the networking business into three areas: scale-up, scale-out, and scale-across. Scale-up links multiple server blades inside a large server rack. Scale-out connects multiple racks and systems into a large cluster, so firms can run a data center like one high-powered computer. Scale-across connects whole data centers to each other, so an AI computer can run across multiple buildings.
To support that, Nvidia uses NVLink, InfiniBand, and Spectrum-X. NVLink connects chips and server blades inside a rack. InfiniBand and Spectrum-X connect clusters and racks, including CPUs, compute nodes, and networking gear. Spectrum-XGS connects multiple data centers.
In December, the company said Amazon’s AWS would pair its custom Trainium chips with Nvidia’s NVLink Fusion, connecting Amazon processors into racks that function as individual computers.
Jensen also said, “Momentum is strong with our Spectrum-X ethernet scale-up and scale-across networking as customers work to unify distributed data centers into integrated, giga-scale AI factories.”
He added, “For the full year, our networking business exceeded $31 billion in revenue, up more than 10x compared to fiscal 2021. Our demand profile is broad, diverse, and expanding beyond just chat bots.”
