The AI Chip Wars Rage On; The Rise of the AI Academic-Turned-Founder; Has No One Seen Black Mirror?

When Microsoft took an early bet on OpenAI in 2019, investing $1 billion into the startup, it also placed a major bet on another project: creating a competitor to Nvidia’s graphics processing units, we reported back in April. Now, years later, Microsoft executives plan to unveil a chip designed for artificial intelligence next month at Ignite, its annual developers conference, we reported last Friday.

To be fair, Microsoft is clearly playing catch up to Google and Amazon, which already sell their own in-house AI chips to customers, as well as Nvidia GPUs. There’s reason, though, to believe Microsoft isn’t that far behind.

So far, the cloud providers’ AI chips haven’t caught on with developers, in part because of the mature software ecosystem around Nvidia GPUs. Amazon just days ago inked a deal with Anthropic aimed at improving its in-house AI chips, after previously turning a similar deal down years ago. Microsoft, on the other hand, has been working closely with OpenAI since 2019, which could give it an advantage in understanding how to design its chips for training and powering large language models.