As Nvidia announced record data center revenue in its quarterly financial results this week, it has been hailed in the financial press as a new “king of chips.” In an interview that founder and CEO Jensen Huang gave to the UK’s Financial Times
, he highlighted the company’s plans to rule the data center. Marrying this up with rumors that Nvidia is considering purchasing Arm, could we be seeing signs of a transition in both the direction and fortune of both Nvidia and Arm?
Record data center revenue
As far as Nvidia’s vision for data center dominance goes, its second quarter of 2021 results demonstrate its goals are set on a sturdy foundation. Revenue of $3.87 billion was up 50% on the year before; that included record data center revenue of $1.75 billion, a huge 167% jump from the previous year, with about 14% of the data center revenue coming from recent acquisition Mellanox.
Huang said, “Adoption of Nvidia computing is accelerating, driving record revenue and exceptional growth. Our new Ampere GPU architecture is sprinting out of the blocks, with the world’s top cloud service providers and server makers moving quickly to offer Nvidia accelerated computing. Mellanox grew sharply, driven by the need for high-speed networking in cloud data centers to scale-out AI services.”
Speaking to analysts, Nvidia’s executive vice president and chief financial officer, Colette Kress, said, “The first Ampere GPU, A100, has been widely adopted by all major server vendors and cloud service providers. Google Cloud platform was the first cloud customer to bring it to market, making it the fastest GPU to come to the cloud in our history.” She added, “Mellanox revenue growth accelerated with strength across Ethernet and InfiniBand products. Our Ethernet shipments reached a new record. Major hyperscale build drove the upside in the quarter as growth in cloud computing and AI is fueling increased demand for high-performance networking. Mellanox networking was a critical part of several of our major new product introductions this quarter.”
[caption id="attachment_7385" align="alignleft" width="250"] Jensen Huang (Image: Nvidia)
In the analyst call, Huang commented on the structural changes in the data center coupled with a new set of dynamics in the need for acceleration of artificial intelligence, the most important application being offered by cloud service providers.
He said the there’s an architectural change in the data center, from hosting applications to hosting services, which he called disaggregated versus hyper-converged. “The original name of hyperscalers is a large data center of a whole bunch of hyper-converged computers. But the computers of today are really disaggregated. A single application service could be running on multiple servers at the same time, which generates a ton of east-west traffic, and a lot of it is artificial intelligence neural network models. And so, because of this type of architecture, two components, two types of technologies are really important to the future of cloud.”
Those two technologies are GPUs and high-speed networking, due to the disaggregated server. “The application is fractionalized and broken up into a bunch of small pieces that are running across the data center. And whenever an application needs to send parts of the answer to another server for the microservice to run, that transition is called east-west traffic. As a result, he said, “The most important thing you could possibly do for yourself is to buy really high-speed, low-latency networking. And that’s what Mellanox is fantastic at. And so, we find ourselves really in this perfect condition where the future is going to be more virtual, more digital.”
Are you buying Arm? No comment, but it seems ideal for data centers
With the current portfolio, Nvidia has the GPU and the networking. So, what’s the next step? In his interview with the Financial Times, Huang said Nvidia planned to supply the full technology stack needed to run data centers. He repeated what he told analysts, that the Arm architecture was ideal, and implied its suitability without being drawn into the “Are you buying Arm?” question. He commented, “We’ve been a long-term partner of Arm, and we use Arm in a whole bunch of applications, whether it’s autonomous driving or a robotics application, or the Nintendo Switch console business that we’re in. Recently, we brought CUDA to Arm to bring accelerated computing to Arm. And so, we worked with the Arm team very closely.”
He added, “They’re really great guys. One of the special [things] about the Arm architecture is that it’s incredibly energy-efficient. And because it’s energy-efficient, it has the headroom to scale into very high-performance levels over time. And so, anyways, we love working with the Arm guys.”
What would happen if Nvidia did buy Arm?
There has of course been a lot of speculation about Arm’s fate, given that the current owner, Softbank has indicated its intention to sell the company it bought for $32 billion back in 2016. As we reported last month
, Softbank has been entertaining offers from potential buyers, as well as considering an IPO. Since then, Nvidia has been closely linked as the potential buyer.
When Nvidia was first named as the potential acquirer, I thought it was potentially reckless, as Bolaji Ojo wrote earlier this month
. Of course, it would alienate some of Arm’s existing licensees. But the more we hear Jensen Huang talk about owning the whole technology stack for the data center, and not directly denying questions about Arm, it does make you think about how this would work.
Before looking at that, what are the other options? Well, an acquisition by any chip company is just going to antagonize some others, just because the world has been Arm centric for some 25 years and even more so now. Almost all the major contenders are likely to be Arm licensees. So that leaves the electronic design automation (EDA) companies, Cadence, Mentor (Siemens) and Synopsys. It would be a no-brainer for one of them, just to add Arm as they already have acquired others like Arc and Tensilica. The only problem is most would not be able to afford the price tag. However, there is the option to team up with a private equity group like the Carlyle Group or KKR.
A more viable option to maintain status quo for Arm and the industry is an IPO. But while viable, it may not be desirable. As one person commented to me this week, “An IPO at $32 billion would not see a huge return on investment for Softbank. An IPO would be a one-way ticket to mediocrity.” The latter was a comment referring to the need to perform the quarterly results cycle, which Arm had already experienced previously as a public company.
An acquisition by Nvidia, though potentially initially disruptive for some industry players, would give Nvidia what it needs to be ‘king of chips’ as one newspaper said this week, giving Nvidia the whole stack, the GPU, CPU, networking, and the software. Nvidia would get all the ace cards in so many markets. As I was told by one executive, “It would give the West – the U.S. and Europe – what they want and more time to breathe, keeping it out of China’s hands. It gives Nvidia about two years to dominate the market, before China catches up with its own competitive CPU.”
If Nvidia did buy Arm, there could certainly be a number of unhappy Arm licensees. But with many currently dependent on Arm, the challenge is not unsurmountable and I am sure Nvidia and Arm could work out engagement models going forward, allowing people to get comfortable with a ‘new norm’, or accelerating their plans with new designs that may not be dependent entirely on Arm.
The deal is not a foregone conclusion though – it’s just that the data center synergy seems to align between the two companies at this moment in time. There are other possible acquirors – like Intel and AMD. But maybe Intel is just sitting tight at the moment and playing a ‘wait and watch’ game, waiting for Nvidia to make its bid, if that is on the cards. It could make a counter-bid at the last moment, which would please Softbank no end if there was a price-bidding war. Or Intel could play the long game, waiting for a possible Nvidia-Arm deal to settle, and then – make a bid on Nvidia. That could then place it back at the top.
It certainly is an interesting time for big-ticket mergers and acquisitions in the electronics industry. And Arm will end up surely being a big coup for anyone putting their money on the table.