Startups like Cerebras, Graphcore, SambaNova, Tenstorrent, Intel Habana etc. have been working for 3+ years to get chips to the market. AFAIK, none had a large success other than small deployments. Habana has a non-public instance type in AWS. Now the easy VC money is gone due to the Fed. The startups have not yet been able to develop an alternative to Nvidia. Are they going to go under? Will they get bought? Will they IPO? Or will VCs continue to fund them?
Heard AMD is also doing some stuff .. but really now it’s almost impossible to replace NVIDIA’s monopoly as for years everyone’s gotten used to their SW stacks
All of the people saying “they will go under” are the ones missed the boat and now want them to fail. Poor tech people!!!
Rather could you explain why they would succeed? It’s a startup, I don’t know what boat you are talking about?
Missed the boat obviously means when they were 10-50M valuation to now 4-5B. Even if they exit at 2-3B, employees make good money so there is nothing like going under. And if they are able to win one big customer their valuation will jump further
Habana is doing better than you think.
Most will go under, especially given the economic downturn. Graphcore seems to be struggling. A lot of other companies are probably in a similar state. It’s tough to get real revenue and beat Nvidia. https://www.thetimes.co.uk/article/f1c7282a-4637-11ed-a9bf-432f673b24aa?shareToken=a742581835c0961b967c5e5484edaa84 https://twitter.com/sumitgup/status/1579496396893622274?s=46&t=aK07QHeTYvAwxMYm9LfDpg
Well graphcore is going under it seems
Graphcore, Cerebras and Sambanova will all go down, already on the way. Don’t know about Tenstorrent. Habana will be canned shortly, Intel is an x86 company again with Pat. Plus manufacturing.
Intel ship has already sunk 😂
Imo Cerebras has the best chance because the technology is very unique that fundamentally sets them apart from others.
Cerebras chips is more like science experiment. Most of the customers aren’t buying more of their hardware or renewing their contracts. Sambanova seems to be better of the lot.
Which one of these is in good shape with the most potential? Which one is likely to fail?