Nvidia has been the artificial intelligence (AI) chip leader since the earliest days of this technology boom, but it now faces threats from unlikely places – its own customers
Google, Amazon, Microsoft, Meta and OpenAI are now developing their own custom AI chips to reduce their dependence on Nvidia’s powerful graphics processing units (GPUs).
Google has developed its Tensor Processing Unit, specifically designed for AI workloads. Amazon has brought its own Trainium chips. OpenAI has also recently introduced its custom AI processor, Jalapeno, which the company claims performed better than Nvidia Corp.’s current lineup in testing. At the same time, Microsoft and Meta are believed to be developing their own accelerators.
The phenomenon could eventually weaken Nvidia’s status as the leader in the AI chip market. But for now, it’s all theory. As of today, Nvidia seems to be getting stronger rather than weaker. And that is because AI demand is growing even faster than the shift towards custom chips.
The chipmaker’s revenue for the July quarter more than doubled to $96.2 billion, while adjusted profit came in at $2.22 per share. The momentum is expected to remain strong over the next year, with the company targeting a 70% increase in revenue in 2028.
Why, then, aren’t these efforts hurting Nvidia already?
One reason is that developing a chip is only part of the challenge. Nvidia has spent years building a complete ecosystem.
For example, Amazon Web Services (AWS), Amazon’s cloud computing arm (which is also developing its own chip), has announced deepening its partnership with Nvidia and plans to buy 2 million more Nvidia GPUs over the next two years.
Nvidia CEO Jensen Huang told CNBC that the deal could also include the purchase of “millions of CPUs.”
Explaining how CPU works, Motley Fool said, “These are central processing units, the main chip in most computers, and are seen as a key chip involved in powering agentic AI.”
Though Nvidia is relatively new to the standalone CPU market, it is making a big push with its Rubin platform. And it expects the standalone CPU business to generate $20 billion in revenue this year and aims to become a major player in the market.
So, in terms of innovation, it is moving much faster than its competitors.
Another factor likely to work in Nvidia’s favour is the sheer scale of the AI infrastructure boom.
Even if Google, Amazon, or OpenAI shift some workloads to their own chips, Nvidia will continue to grow if overall demand for AI computing expands rapidly enough.
For now, however, Nvidia appears to have bought itself time.
That doesn’t mean the threat has disappeared
The threat actually comes from rising costs.
“Memory chips have become much more expensive because of the AI boom. As a result, Nvidia’s profit margin is expected to fall from around 75% to 71–72% by January,” says Viram Shah, Founder & CEO of Vested Finance.
Also, Nvidia is locking in more supplies to avoid shortages. Its supply commitments have jumped from $119 billion to $279 billion in just three months, largely to secure sufficient memory.
And now to deal with the higher costs, Nvidia will have to raise prices next year. “The big question is whether customers will accept the higher prices or push back. Their response will have a major impact on Nvidia’s profit margins next year.”