Intel Is $2B Serious About AI, Buys Habana to Target Nvidia

Intel is betting on a $2 billion acquisition of artificial intelligence (AI)-focused startup Habana Labs to bolster its position in the highly competitive data center processor space that the chip giant predicts will be worth more than $25 billion by 2024. Analysts noted the move could also highlight issues with its current Nervana platform. Intel is betting on a $2 billion acquisition of artificial intelligence (AI)-focused startup Habana Labs to bolster its position in the highly competitive data center processor space that the chip giant predicts will be worth more than $25 billion by 2024. Analysts noted the move could also highlight issues with its current Nervana platform. The deal will see Intel fork over a significant multiple for Israel-based Habana Labs. Intel’s venture capital arm Intel Capital was one of four investors in Habana, including the chip company’s most recent $75 million Series B round late last year. But the Crunchbase report also noted that Habana generated just $2 million in annual revenues. Habana’s current portfolio is based on two platforms. First is its Goya AI Inference Processor, which it launched late last year, and is being used by Facebook for its Glow machine learning (ML) compiler. The second is its Gaudi Training Processor that it launched in June. “This acquisition advances our AI strategy, which is to provide customers with solutions to fit every performance need – from the intelligent edge to the data center,” said Navin Shenoy, EVP and GM of Intel’s Data Platforms Group. “More specifically, Habana turbocharges our AI offerings for the data center with a high-performance training processor family and a standards-based programming environment to address evolving AI workloads.” Alan Priestley, VP and analyst at Gartner, explained in an email to SDxCentral that the move should expand Intel’s ability to compete in the broad AI space. “AI is a very broad spectrum of techniques and algorithms and cannot efficiently be addressed by a single chip architecture, so having multiple architectures will provide Intel wider market coverage,” Priestley wrote. “Also if they are able to integrate Habana’s software sack into their OneAPI initiative this will simplify software development while providing choice in execution hardware.” Analysts noted the deal should better position Intel against rivals like Nvidia, which Karl Freund, senior analyst for high-performance computing (HPC) and deep learning at Moor Insights & Strategy, noted in a Forbes post was the “800-pound gorilla in this space.” “Clearly, Intel realizes that it needs breakthrough performance and efficiency to go up against Nvidia,” Freund wrote. Habana’s Chief Business Officer Eitan Medina previously told SDxCentral that the firm has targeted Nvidia’s market dominance. He explained that benchmark tests showed that both its Goya and Gaudi platforms outperformed Nvidia GPUs and Intel CPUs. “Frankly, our customers are hungry for an alternative even if [our processors] weren’t so much better,” Medina said. “Our motto is AI performance, not stories. We are talking about actual application performance, and we are giving you hardware with software that you can test.” Habana is set to remain an independent business unit within Intel’s Data Platforms Group, maintain its current management team, and will continue to be based in Israel. Freund did note it was interesting that Habana would be reporting into Shenoy and not Naveen Rao, who was former CEO of Nervana and head of its AI products. The move also shines new light on Intel’s Nervana Neural Network Processors plans that it announced with Facebook earlier this year. That platform is a new class of chip designed for AI inference workloads, which is when a machine acts on a new data sample to infer an answer to a query. “This is a really big deal for us,” Shenoy said earlier this year on the launch. “It expands our position in AI above and beyond what we’ve done in Xeon and in Core into a new domain.” However, Freund explained that Nervana “uses a proprietary interconnect for scaling, while Habana’s Gaudi can scale to thousands of nodes over standard 100Gb Ethernet [GbE].” Shenoy noted that Habana will bring those advances. “We know that customers are looking for ease of programmability with purpose-built AI solutions, as well as superior, scalable performance on a wide variety of workloads and neural network topologies,” he stated. Intel announced just last month that its Nervana processors were in production and being delivered to customers. It said at that time that its AI portfolio was on target to generate $3.5 billion in revenue this year. Freund stated that Intel “can’t be happy with its Nervana efforts to date. Intel must get this right; I don’t believe it will get a third chance, but it is still early enough in the game to switch horses.”