Investment Rush Against Nvidia

Funding is pouring into AI inference chip startups, rapidly growing the anti-Nvidia ecosystem.

According to Dealroom, global investment in AI chip startups reached $8.3 billion in 2026, and further expansion is expected this year as well.

The key is dedicated architectures optimized for the deployment and inference stage rather than training GPUs. Founders argue that existing GPUs are not optimized for large-scale inference, and new designs can create power and cost savings.

Major funding cases are as follows.

• United States: Cerebras Systems $1 billion, MatX $500 million, Ayar Labs $500 million, Etched $500 million
• Europe: Axelera, Olix, Vaire Computing, Fractile, and others raised rounds of $200-300 million or more

Nvidia is also responding, reportedly investing more than $18 billion in R&D in its most recent fiscal year.

Ultimately, the inference chip competition is being read as a signal that will lower AI infrastructure costs and accelerate the multipolarization of the hardware market.