- Investor Michael Burry has compared Nvidia's situation to Cisco's historical reliance on SmartNet support agreements.
- He noted that Cisco maintained dominance for about a decade but suffered a decline in its stock price after the tech bubble burst.
- He claimed that Nvidia’s position could be threatened as major tech firms increasingly invest in their own custom AI chips.
Investor Michael Burry has raised concerns that Nvidia's future might resemble Cisco Systems' after the dot-com boom. He says that the chipmaker's perceived competitive moat may not be as strong as many investors think.
Burry wrote an article on Sunday comparing Nvidia to Cisco in the early 2000s. He said that Cisco was able to stay on top because its SmartNet support agreements kept customers locked in. These contracts made clients buy ongoing support services, which made it hard for them to switch because of the costs of certifications, maintenance, and infrastructure dependencies.
Burry, who’s often called The Big Short, claimed this layer of contracts helped Cisco stay in business even when new competitors came along. The company stayed on top for about 10 years and made a lot of money in the years that followed, but its stock price fell after the tech bubble burst and took years to return to its previous level.
Burry said that Nvidia's ecosystem, especially its CUDA software platform, is similar to Cisco's historical advantage, but it doesn't have the same contractual lock-in. CUDA helps developers make software run better on Nvidia GPUs, making the AI ecosystem a bit dependent on it. He did say, though, that the platform isn't legally tied to Nvidia hardware.
Nvidia Corp. (NVDA) was in the red in after-hours trading. On Stocktwits, retail sentiment around NVDA remained in ‘bearish’ territory, as chatter levels remained in ‘low’ over the past day.
NVIDIA’s CUDA Might Not Give It The Competitive Advantage
He further stated that several semiconductor companies are developing chips compatible with CUDA, a Nvidia-backed computing platform.
Burry claimed this could let developers continue using the software ecosystem while running workloads on hardware other than Nvidia's. He argued that if this kind of compatibility becomes common, CUDA's ability to act as a long-term competitive moat could weaken.
In an effort to reduce dependence on Nvidia's hardware, big tech companies are increasingly developing their own custom AI chips. Burry pointed out that companies like Amazon (AMZN), Google (GOOG), and Apple (AAPL) are making significant investments in internal processors like Trainium and Tensor Processing Units (TPUs).
As an example of how big tech companies are already experimenting with alternatives to Nvidia's GPUs, Burry pointed out that AI startup Anthropic trains models using Trainium and Google TPUs.
Last year, Anthropic implemented a multi-platform computation strategy, training and deploying its Claude models on Google's TPUs, Amazon's Trainium CPUs, and Nvidia GPUs.
Read also: Bitcoin Defies Geopolitical Shock As Iran Threatens Ukraine, Analyst Signals A ‘Regime Shift’
For updates and corrections, email newsroom[at]stocktwits[dot]com.