NVIDIA has responded to all the 'buzz' around Google's TPUs and reports of it being externally adopted, as the firm says that ASICs are limited to a "specific AI frameworks".
NVIDIA Says ASICs Are Limited To a Specific Workload, While the Firm Is Responsible For the 'Entire' AI Revolution
Google TPUs have been the 'talk of the town' in recent times, especially since there have been reports that the AI chips from the company are being adopted externally, from the likes of Meta and Anthropic. A narrative has emerged, claiming that an ASIC manufacturer is poised to replace NVIDIA in a segment dominated by Team Green for several years now. Building upon this, NVIDIA has responded to reports about Google's TPUs, saying it is "delighted" by the success of the tech giant. However, at the same time, the firm has responded to the competition with ASICs.
We’re delighted by Google’s success – they’ve made great advances in AI, and we continue to supply to Google. NVIDIA is a generation ahead of the industry – it’s the only platform that runs every AI model and does it everywhere computing is done.
NVIDIA offers greater performance, versatility, and fungibility than ASICs, which are designed for specific AI frameworks or functions.
- NVIDIA's spokesperson in a statement to Wccftech
The statement comes from an earlier report by The Information, which claimed that Meta is in line to purchase 'billions of dollars' worth of Google TPUs for their AI workloads, and eventually, it was projected that the external adoption of Google's ASICs could account for 10% of NVIDIA's AI revenue. The idea here is that Google has successfully vertically integrated its AI workloads with self-built TPUs, particularly in inference workloads, and has achieved superior performance parameters compared to what NVIDIA offers.
It won't be wrong to say that, out of all ASIC pursuants, Google is one of the most competitive ones, especially since the firm has been in the game for almost a decade now. But, based on what NVIDIA believes, ASICs are designed for "specific frameworks", while the company's tech stack, whether it is the computing architecture or the CUDA platform, targets the whole AI ecosystem. Additionally, it is worth noting that Google is a significant customer of NVIDIA's AI hardware, making TPUs a key component of a broader market where NVIDIA remains in the leading position.
It would be interesting to see how the race between ASICs and NVIDIA's technology progresses, but the segment will certainly become significantly more competitive, especially as we move into a world where inference is the 'real deal' for AI giants.
Follow Wccftech on Google to get more of our news coverage in your feeds.
