Nvidia Custom Chip Competition Threat
Broadcom's custom chip order and Google's TPUs pose key risks to Nvidia's market position.
Too little corroboration in the last 3 days to call a trend (10 articles). Watching for it to gain traction.
Mainstream financial press is carrying this — attention has broadened beyond specialist outlets.
"Google (GOOG) (GOOGL) TPUs, developed with Broadcom, represent a custom-accelerator alternative to Nvidia's GPUs. Broadcom is supplying the silicon and networking while simultaneously helping create the financial infrastructure that gets those systems into AI data centers."
"Micron: It Does Not Need To Beat SK hynix At Nvidia... Micron Technology (MU) appears to be judged on HBM with the question: can Micron take enough NVIDIA (NVDA) business from SK hynix (SKHY)? But this is not the right"
"Google has been pitching its Tensor Processing Units to so-called neoclouds—specialist cloud firms that rent out AI computing power—in a bid to break into a market Nvidia has owned almost from end to end. It already counts Anthropic, Meta and Apple among TPU users, and has struck a $5 billion deal with Blackstone to build a TPU-based neocloud that starts renting capacity next year."
"The mounting strain on the firm's product lines highlights concerns that Nvidia's breakneck annual product launches are colliding with manufacturing limits. This gives rivals AMD and Google a chance to capture a dominant share of the market."
"But Broadcom is not trying to beat Nvidia on Nvidia's terms. It is competing in a different lane — custom silicon for production workloads at hyperscaler scale."
"It's improving ads, search, and cloud growth, while also building out its own infrastructure. And it's reducing reliance on Nvidia with custom chips."
"The move reflects a dramatic shift in sentiment as Alphabet emerges as both a major AI services provider with its cloud platform and a key rival to Nvidia in chips through custom processors that have won customers such as Anthropic."
"Google and Amazon (AMZN) are already developing their own chips. Every major Nvidia customer that goes in-house reduces the addressable market Huang is counting on to hit that $1 trillion target."
"Demand for custom silicon like TPUs has surged recently. Hyperscalers and startups are seeking efficient alternatives to Nvidia’s (NVDA) costly GPUs (graphics processing units)."
"The main risk with the trade is a sharp move lower early in the trade."