[1/3]A detailed-up view of the probe station that’s getting used to check the Azure Cobalt chip at Microsoft’s lab in Redmond, Washington is seen on this undated handout photograph obtained by Reuters on November 13, 2023. Courtesy of Microsoft/Handout through REUTERS Purchase Licensing Rights
Nov 15 (Reuters) – Microsoft (MSFT.O) on Wednesday introduced a duo of custom-designed computing chips, becoming a member of different huge tech corporations that – confronted with the excessive value of delivering synthetic intelligence providers – are bringing key applied sciences in-house.
Microsoft mentioned it doesn’t plan to promote the chips however as an alternative will use them to energy its personal subscription software program choices and as a part of its Azure cloud computing service.
At its Ignite developer convention in Seattle, Microsoft launched a brand new chip, known as Maia, to hurry up AI computing duties and supply a basis for its $30-a-month “Copilot” service for enterprise software program customers, in addition to for builders who wish to make {custom} AI providers.
The Maia chip was designed to run massive language fashions, a sort of AI software program that underpins Microsoft’s Azure OpenAI service and is a product of Microsoft’s collaboration with ChatGPT creator OpenAI.
Microsoft and different tech giants corresponding to Alphabet (GOOGL.O) are grappling with the excessive value of delivering AI providers, which might be 10 occasions higher than for conventional providers corresponding to engines like google.
Microsoft executives have mentioned they plan to sort out these prices by routing practically the entire firm’s sprawling efforts to place AI in its merchandise by means of a standard set of foundational AI fashions. The Maia chip, they mentioned, is optimized for that work.
“We expect this offers us a method that we are able to present higher options to our clients which can be quicker and decrease value and better high quality,” mentioned Scott Guthrie, the manager vice chairman of Microsoft’s cloud and AI group.
Microsoft additionally mentioned that subsequent yr it is going to provide its Azure clients cloud providers that run on the latest flagship chips from Nvidia (NVDA.O) and Superior Micro Units (AMD.O). Microsoft mentioned it’s testing GPT 4 – OpenAI’s most superior mannequin – on AMD’s chips.
“This isn’t one thing that is displacing Nvidia,” mentioned Ben Bajarin, chief govt of analyst agency Inventive Methods.
He mentioned the Maia chip would permit Microsoft to promote AI providers within the cloud till private computer systems and telephones are highly effective sufficient to deal with them.
“Microsoft has a really totally different form of core alternative right here as a result of they’re making some huge cash per consumer for the providers,” Bajarin mentioned.
Microsoft’s second chip introduced Tuesday is designed to be each an inside value saver and a solution to Microsoft’s chief cloud rival, Amazon Internet Providers.
Named Cobalt, the brand new chip is a central processing unit (CPU) made with know-how from Arm Holdings (O9Ty.F). Microsoft disclosed on Wednesday that it has already been testing Cobalt to energy Groups, its enterprise messaging software.
However Microsoft’s Guthrie mentioned his firm additionally needs to promote direct entry to Cobalt to compete with the “Graviton” sequence of in-house chips supplied by Amazon Internet Providers (AWS) (AMZN.O).
“We’re designing our Cobalt answer to make sure that we’re very aggressive each when it comes to efficiency in addition to price-to-performance (in contrast with Amazon’s chips),” Guthrie mentioned.
AWS will maintain its personal developer convention later this month, and a spokesman mentioned that its Graviton chip now has 50,000 clients.
“AWS will proceed to innovate to ship future generations of AWS-designed chips to ship even higher price-performance for no matter buyer workloads require,” the spokesman mentioned after Microsoft introduced its chip.
Microsoft gave few technical particulars that may permit gauging the chips’ competitiveness versus these of conventional chipmakers. Rani Borkar, company vice chairman for Azure {hardware} methods and infrastructure, mentioned each are made with 5-nanometer manufacturing know-how from Taiwan Semiconductor Manufacturing Co (2330.TW).
She added that the Maia chip can be strung along with customary Ethernet community cabling, relatively than a dearer {custom} Nvidia networking know-how that Microsoft used within the supercomputers it constructed for OpenAI.
“You will note us going much more the standardization route,” Borkar advised Reuters.
(This story has been refiled to appropriate Rani Borkar’s gender to feminine, not male, in paragraph 19)
Reporting by Stephen Nellis in San Francisco; Modifying by Edmund Klamann and Jonathan Oatis
Our Requirements: The Thomson Reuters Belief Rules.
