Nvidia stock slipped on Wednesday as investors reacted to fresh competitive pressure from Amazon’s new Trainium 3 artificial intelligence chip, the latest sign that major cloud providers are accelerating efforts to develop their own AI silicon.
At the time of publishing, the Nvidia stock was down 0.6% to trade at around $180.34.
Amazon unveiled Trainium 3 on Tuesday, pitching it as a cost-efficient alternative for training and operating AI models.
The company said the new chip can reduce AI training and inference costs by up to 50% compared with systems using equivalent GPUs — the category dominated by Nvidia.
Amazon also said it plans to use Nvidia’s NVLink Fusion technology in its future AI computing infrastructure, integrating it with the forthcoming Trainium4 chip.
“With Nvidia NVLink Fusion coming to AWS Trainium4, we’re unifying our scale-up architecture with AWS’s custom silicon to build a new generation of accelerated platforms,” Nvidia CEO Jensen Huang said.
“Together, NVIDIA and AWS are creating the compute fabric for the AI industrial revolution.”
Nvidia stresses long-term demand despite competitive moves
Nvidia is working to reassure investors that it can maintain dominant market share even as Amazon, Google and other hyperscalers expand use of in-house silicon.
The company’s neutral position in the market — as a supplier rather than a direct cloud-services competitor — remains a strategic advantage, as some technology giants may prefer not to depend heavily on rival hardware.
Nvidia CFO Colette Kress said Tuesday that AI models trained on its new Blackwell chips will begin emerging in about six months.
She noted the company has $500 billion in bookings for Blackwell and Rubin chips through 2026, excluding an upcoming deal with OpenAI that has yet to be finalised.
Separately, European AI start-up Mistral said it trained its next-generation models on Nvidia hardware.
The companies highlighted that Mistral’s Large 3 model achieved a tenfold performance improvement on Nvidia’s GB200 NV72 server racks compared with the previous H200 generation.
Competition landscape starting to get intense
While Oracle Cloud Infrastructure’s earlier adoption of more than 50,000 AMD chips signalled growing interest in non-Nvidia solutions, the competitive pressure now is coming most visibly from Amazon Web Services.
With Trainium 3, AWS has taken a significant step toward deepening its in-house AI silicon strategy.
The chip is said to offer four times the performance of its predecessor and reduces energy consumption by 40%, underscoring AWS’s ambition to optimise its data centres around its own hardware.
Google, meanwhile, is extending more aggressive outreach for its Tensor Processing Units, promoting TPUs to major customers such as Meta.
The push suggests Google is seeking to expand TPU adoption among hyperscalers that have traditionally relied on Nvidia GPUs.
The combined efforts of Amazon, Google and AMD signal a broadening competitive landscape in the AI hardware sector.
While Nvidia remains the clear leader, its largest customers are now among its most visible challengers — each moving to reduce reliance on external suppliers and expand control over their AI infrastructure.
The post Nvidia stock continues slide: is the AI darling’s moat drying up as competition intensifies? appeared first on Invezz
