Microsoft AI Watch posted an update
Microsoft plans to scale its Maia chips to multi-gigawatt deployments from next year, with Maia 300 claimed to deliver 40% more tokens per watt than leading GPU systems. The announcement, described by Yahoo Finance, puts Microsoft’s in-house AI silicon on a much larger footing.
Why it mattersBloomberg Intelligence analyst Mandeep Singh argues Microsoft remains well behind Google’s TPUs and Nvidia’s GPUs. That is an analyst’s assessment, not a like-for-like benchmark in the supplied report. For Microsoft, the test is whether Maia can take a meaningful share of inference workloads and chip spending, rather than merely join the fleet.
Discuss: Should Microsoft prioritise scaling its own Maia chips even if it still needs Nvidia and other suppliers, or is that a costly halfway house?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.