[ad_1]
At its annual developer convention, Ignite, Microsoft on Wednesday unveiled the long-anticipated {custom} cloud computing chip for its Azure cloud service, referred to as Azure Maia 100, which it stated is optimized for duties akin to generative AI.
The Maia 100 is the primary in a sequence of Maia accelerators for AI, the corporate stated. With 105 billion transistors, it’s “one of many largest chips on 5-nanometer course of know-how,” stated Microsoft, referring to the dimensions of the smallest options of the chip, 5 billionths of a meter.
Additionally: Microsoft’s newest AI choices for builders revealed at Ignite 2023
As well as, the corporate launched its first microprocessor constructed in-house for cloud computing, the Azure Cobalt 100. Like Maia, the processor is the primary in a deliberate sequence of microprocessors. It’s primarily based on the ARM instruction-set structure from ARM Holdings that’s licensed to be used by quite a few firms together with Nvidia and Apple.
Microsoft stated Cobalt 100 is a 64-bit processor that has 128 computing cores on die, and that it achieves a 40% discount in energy consumption in comparison with different ARM-based chips that Azure has been utilizing. The Cobalt half is already powering packages together with Microsoft Groups and Azure SQL, stated the corporate.
The 2 chips, Maia 100 and Cobalt 100, are fed by 200 gigabit-per-second networking, stated Microsoft, and may ship 12.5 gigabytes per second of knowledge throughput.
Microsoft is the final of the Large Three cloud distributors to supply {custom} silicon for cloud and AI. Google pioneered the race to {custom} silicon with its Tensor Processing Unit, or TPU, in 2016. Amazon adopted swimsuit with a slew of chips together with Graviton, Trainium, and Inferentia.
Rumors of Microsoft’s efforts have circulated for years, fed by occasional disclosures akin to final summer time’s leak of a planning doc from the corporate.
Additionally: Azure AI Studio takes the stage at Ignite 2023: Unlock the potential of this AI toolkit
Microsoft made some extent of noting that it continues to associate with each Nvidia and AMD for chips for Azure. It plans so as to add Nvidia’s newest “Hopper” GPU chip, the H200, subsequent 12 months, in addition to AMD’s competing GPU, the MI300.
Microsoft’s chips will help with packages akin to GitHub Copilot, however they may also be used to run generative AI from AI startup OpenAI, into which Microsoft has poured $11 billion in funding to safe unique rights to packages akin to ChatGPT and GPT-4.
At OpenAI’s developer convention final week in San Francisco, Microsoft CEO Satya Nadella pledged to construct “the perfect compute” for OpenAI “as you aggressively push ahead in your roadmap.”
Additionally: OpenAI unveils {custom} GPTs, Retailer, and Assistant API, at Dev Day
Microsoft and OpenAI are each attempting concurrently to lure enterprises to make use of generative AI. Microsoft is seeing massive progress within the generative AI enterprise, Nadella informed Wall Road final month. The corporate’s paying prospects for its GitHub Copilot software program rose by 40% within the September quarter from the prior quarter.
“We now have over 1 million paid Copilot customers in additional than 37,000 organizations that subscribe to Copilot for enterprise,” stated Nadella, “with vital traction outdoors the US.”
Additionally at Ignite, Microsoft introduced it’s extending Copilot to Azure with a public preview of Copilot for Azure, a device it stated will give system directors an “AI companion” that can “assist generate deep insights immediately.”
Additionally: OpenAI CEO: We’re completely satisfied if Microsoft makes a sale, and so they’re completely satisfied if we make a sale
Along with the chip improvements, Microsoft introduced basic availability of Oracle’s database packages operating on Oracle {hardware} within the US East Azure area. Microsoft is the one cloud operator to supply Oracle database on Oracle’s personal laptop methods infrastructure, it stated.
Different associate information included the final availability of Microsoft’s edge computing service, Arc, for VMware’s vSphere virtualization suite.
[ad_2]
Source link