SoftBank to Purchase Silicon Valley Chip Start Up Ampere for $6.5 billion
SoftBank announced on Wednesday that it has agreed to pay $6.5 Billion for Silicon Valley chip startup Ampere Computing. The deal reflects the Japanese conglomerate’s belief that Ampere’s chips can play a significant part in artificial intelligence where Nvidia is reaping the most rewards. Ampere was founded in 2008 to sell data center chips based on technology licensed from Arm Holdings. This British company licenses chip designs which have powered almost all mobile phones. SoftBank, who bought Arm in 2016, is working to make chips based on Arm Technology more widely used for different tasks. Ampere’s expertise will accelerate this vision and deepen our commitment to A.I. innovation in the United States.”
SoftBank said it would operate Ampere as a wholly owned subsidiary under its own name.
The sale comes amid a flurry of deals and shifting alliances driven by a furious demand for the chips used to power A.I. Applications such as OpenAI ChatGPT are a good example. SoftBank has announced a number of transactions to increase its role in this field. Son, OpenAI’s CEO, Sam Altman and Larry Ellison of Oracle, Ampere’s biggest investor and customer, announced Stargate with President Trump at the White House in January. Son, Mr. Altman, and Mr. Ellison stated that Stargate would invest up to $500 billion in building a network of U.S. Data Centers to power OpenAI’s operations, starting with one location in Texas. Nvidia is listed as the key technology partner of the venture. It supplies GPUs (graphics processing units), which are the basis for A.I. calculations.
Another kind of chip also plays central roles in A.I. These are the microprocessors from Intel, Advanced Micro Devices, and Arm that perform general-purpose computing. These chips, also known as “host” processors and working alongside GPUs, manage A.I. Models are used to build special software programs. Typically, one microprocessor is used for every four Nvidia graphics cards sold. Inferencing is a task that includes answering questions in chatbots. Intel and AMD chips have accounted for the majority of A.I. up until now. host processors and microprocessors used for inferencing.
But some influential companies want to change that. Nvidia is pushing Arm processors to replace Intel or AMD chips as host microprocessors. IDC, a market-research firm, predicts the market for microprocessors used for A.I. will grow to $33 billion by 2030 from $12.5 billion in 2025.
A lot of money is at stake. They added that Nvidia was not exclusively backing Arm technology and still supported their chips as an option along with its latest GPUs. They also said that Nvidia wasn’t exclusively supporting Arm technology, and still supported its chips as an optional option along with its newest GPUs. It recently announced plans for an Aurora chip that has up to 512 small computing engines. The company claims this design is best suited for A.I. inferencing applications.
The company, led by Renee James, a former Intel executive, has had some successes. But the biggest spenders in the sector — giant companies like Amazon, Google and Microsoft — have lately focused more on developing their own microprocessors based on Arm technology, rather than relying on the start-up.
Oracle is an exception. Oracle has disclosed its equity and debt investments and has offered online services that are powered by Ampere’s chips. As of May, Oracle said it held a 29 percent stake in Ampere; it put the value of its investments, after accounting for losses, at $1.5 billion.
As part of the purchase, Oracle and Carlyle Group, the big private equity firm that is also a major Ampere investor, agreed to sell their stakes in Ampere, SoftBank said.
Bloomberg reported last month that SoftBank was near a deal to buy Ampere.
