What might have been the largest chip acquisition in history is officially dead. Now Arm, one of the most important companies in the industry, must chart a new path forward.
Late Monday, amid regulatory pressure on three continents, SoftBank and Nvidia said they have abandoned Nvidia’s troubled $40 billion bid for Arm.
The deal was always a long shot for Nvidia: The potential Arm acquisition encountered resistance from U.K. regulators and officials in its native country, but also ran into problems in the EU, and was the target of a Federal Trade Commission antitrust lawsuit in the U.S. China’s approval wasn’t a sure bet either. The deal would have added Arm’s chip designs, which power most of the world’s smartphones, to Nvidia’s assets and potentially given Nvidia a way to even further challenge AMD and Intel in the data center.
But for Arm, the stakes were considerably higher. SoftBank said Monday night that it now plans to take Arm public once again, and will complete the process by the end of March next year. It also said that CEO Simon Segars had resigned, and would be replaced by Rene Haas, who was head of the company’s IP group.
The new direction on which Arm is set to embark was one it insisted it didn’t want to take up until Monday evening. In the year and a half since Nvidia announced the deal, Arm has said in blog posts and regulatory filings that a combination with Nvidia is a better outcome for the company than an initial public offering. The argument advanced by executives was that without Nvidia’s vast resources and ongoing support of developing Arm’s designs, Arm will fall behind and fail to grow into new markets, such as the data center.
Last year, Segars took the argument a step further, and wrote that an IPO would, in fact, damage Arm’s ability to grow and innovate.
“We contemplated an IPO but determined that the pressure to deliver short-term revenue growth and profitability would suffocate our ability to invest, expand, move fast and innovate,” Segars wrote in a blog post last year. “Combining with NVIDIA will give us the scale, resources and agility needed to maximize the opportunities ahead.”
But during a briefing with reporters early Tuesday, Arm executives said the company is enjoying the fruits of a strategy put in place roughly four years ago that involved focusing on more profitable markets and shedding units that operated at a loss. Going public, Haas said, would give Arm more access to cash should it be necessary to fund innovation.
“As I think about the next five to seven years or so, being a public company actually gives us access to capital should we need it,” Haas said. SoftBank disclosed Monday that it expects Arm’s fiscal 2021 adjusted profit to rise to $900 million, on revenue of $2.5 billion, but didn’t disclose the factors it took into account making those adjustments.
Despite Haas’ optimism about Arm’s chances, industry watchers are skeptical about Arm’s ability to fund a big push into the data center without help from Nvidia or by other means. If Arm is banking so much profit this year, it’s unclear why it would also need to raise more cash through an initial public offering, according to Dylan Patel, who founded SemiAnalysis. “Why are you talking about raising capital to fund development if you’re profitable?” Patel said.
What’s clear is that competing with the x86 server chips produced by AMD and Intel has proven costly and difficult. Leading edge node chips are expensive to design, with estimates ranging from $80 million to more than $500 million for a single chip generation. Arm designs made by the likes of AWS and others have made inroads into the data center market but haven’t significantly disrupted AMD and Intel’s duopoly to date.
“If they don’t have a cash infusion, they’re going to find it very hard to compete, in particular, in the data center, the big, big money-maker,” Creative Strategies CEO and tech analyst Ben Bajarin told Protocol. “Anything outside of the [Nvidia] deal will not yield results. An IPO will not yield them a massive amount of cash they can then go and spend on R&D.”
On Nvidia’s part, if it had managed against the odds to close the Arm acquisition, it would have boosted an already successful company through deeper integration of Arm’s tech. But Nvidia can still work closely with Arm’s designers as an architectural licensee.
Without Arm, Nvidia will undoubtedly continue its growing success in data center chips. It already enjoyed a strong partnership with Arm and its data center product strategy — including its Grace server chip and its data processing unit products — is in no danger of being hurt. Baird chips analyst Tristan Gerra wrote in a research note Tuesday that there was no change to his outlook on Nvidia or its tech roadmap.
“We expect Nvidia to continue leveraging its Arm-based architectures, notably with its Bluefield DPU product roadmap, with programmable Arm cores complementing GPU and smart [networking card] architectures, offloading from traditional x86-based configurations and notably serving AI cloud native applications,” he wrote.
Aside from losing out on the possibilities of a deeper collaboration with Arm, the main downside for Nvidia is the additional $1.25 billion breakup fee it now must pay to SoftBank. But Wall Street expects Nvidia to book a $11 billion profit in its current year, according to analyst data from Sentieo.