Arm, Intel Google and Meta launch self-designed AI chips
The future direction for dominance in the AI chip space took another turn this week as four huge stakeholders revealed their own products.
Arm, Intel Google and Meta all announced new products, prompting industry analysts to all ask the same question: what does this mean for Nvidia?
Nvidia is estimated to control more than 90% of the AI chip market, and demand for its market-leading semiconductor designs has propelled the firm to extraordinary growth. Earlier this year, it announced quarterly profits of USD 12.3 billion. If Nvidia’s biggest customers start making their own chip designs, where does this leave the design company. Well, now some of them have – albeit in a limited way. This week saw the following launches:
- Arm revealed its Ethos-U85 NPU, with a 4X performance enhancement and 20% greater power efficiency, as well as 2,048 MAC units.
- Google launched its Axion processor – the company’s first Arm-based CPU for data centres. It comprises the Arm Neoverse V2 CPU.
- Intel unveiled the Gaudi 3 AI accelerator, with a 1.5X increase in memory bandwidth compared to its predecessor.
- Meta launched its latest Training and Inference Accelerator (MTIA) v1 was first announced last year. It said it is three times more efficient than the first version.
All the above firms have experimented with their own designs before, so the announcements are not earth-shattering. But they do reflect the desire of some of the world's biggest firms to assert more control over their own AI platform chipsets.
