Introduction
NVIDIA (NVDA, Financial) is well known for its Graphics Processor Units (GPUs) and Computing & Networking solutions.
The company's history is related to its strong presence in the Graphics solutions market. The recent surge in the computing power requirements for advanced Graphics applications (mainly related to the Gaming market) has been a huge tailwind for NVIDIA.
The other division, Computing & Networking, has a clear target: helping NVIDIA become the leading AI (Artificial Intelligence) solutions company on the market. AI can be used in a wide range of applications, like service robots, self-driving mobility and IoT (Internet-of-things) applications in general.
As an effective AI solution is strongly dependent on computing power, in order to achieve that dominance, they must be able to offer the best suitable computing platform that can be used to run this kind of applications. Providing the AI "platform of choice" will position them to be a fundamental part of the ecosystem, and consequently allow them to reap big returns for decades.
Arm Holdings' Acquisition
NVIDIA's future direction became clear last September, when they announced the intention to acquire private chip designer Arm Holdings from Japanese tech giant Softbank (SFTBY, Financial) in a transaction valued at around $40 billion. Softbank had, in turn, previously acquired Arm Holdings in September 2016 for a total consideration of $31 billion.
Specifically, the company agreed to pay Softbank $12 billion in cash and $21.5 billion in NVIDIA common stock (this will be achieved by issuing 44.3 million shares). They will also issue $1.5 billion in equity to Arm employees. Finally, Softbank may receive up to $5 billion in cash or common stock, subject to the realization of determined performance targets by Arm.
The deal looks like a win/win/win situation for NVIDIA shareholders, who will potentially benefit from the acquisition (despite the announced 7% dilution), Softbank, that will both get cash plus retain a stake in the larger company, and Arm employees, who will enjoy the unexpected bonus.
The acquisition makes total sense from the strategic point of view: NVIDIA will acquire a computing ecosystem leader, with a solid moat, a large market share and precious long-running customer relationships.
Actually, I said "strategic," as this is not just a deal in which a company acquires another company to strengthen its presence in a specific market or acquire a new technology.
There's more than that: NVIDIA could in the future potentially make things much more difficult for its competitors and for other big tech companies (namely, Arm Holdings's current customers) to continue using and leveraging Arm's platform.
As proof of this, last February a consortium of big tech names composed by Alphabet's Google (GOOGL, Financial), Microsoft (MSFT, Financial), Qualcomm (QCOM, Financial) and other players openly opposed the deal, with some of them registering objections to the Federal Trade Commission, the European Commission and other agencies and regulators, claiming that NVIDIA could potentially limit other companies's access to Arm technology or, alternatively, increase prices.
This has never been a problem both before and after Sofbank's acquisition, as Arm has always been recognized for its neutrality: the company offers its products to whoever is willing to pay for them, and has never favored one customer or the other, nor was it involved in competition matters.
On the other side NVIDIA is, of course, promising that nothing will change and that they plan to continue with Arm's open licensing model. They also not only pledged to keep Arm's headquarters in the U.K., but announced their will to expand the Cambridge campus in order to make it a world-class Research Center focused on AI.
Nobody knows if the regulators will give NVIDIA the green light needed to complete the deal, but for sure the closure will be delayed as they now need to provide much more documentation and details about it.
But, even in the case the deal goes through, there's one more thing NVIDIA should be worried about, and that is the rise of RISC-V processors.
RISC-V adoption is a threat for Arm dominance
Before delving into the implication of RISC-V (Reduced Instruction Set Computer) processors adoption, let's see how Arm Holdings's licensing model actually works.
While AMD, Intel (INTC) and NVIDIA itself make money by selling chips, Arm's business model is different, as it is basically an IP (Intellectual Property) licensing business, with a pay-for-play option.
This means that Arm does not sell chips, it actually sells chip designs (while the customer is responsible for implementing and personalizing the chip), and collects:
- An upfront licensing fee, depending on the design complexity
- Per-chip basis royalties
The royalty component can be way bigger than the licensing fee, as it grows with the number of chips sold (it usually amounts to 1-2% of the selling price of the chip).
Source: NVIDIA to acquire Arm Presentation
For years Arm Holdings's customers have paid fees and royalties in order to continue selling their chips, giving away a part of their margins but getting the benefit of having a strong design for their chips, a great support and the possibility to easily customize them.
But what if, after NVIDIA's acquisition, they could not access Arm technology or are forced to pay higher prices which will further eat into their precious margins? With their profits put in jeopardy, they should at least consider the possibility to have a B plan.
And here comes RISC-V, which is not an Arm competitor, and it's also not a product. RISC-V is an open-source microprocessor Instruction Set architecture (ISA), which lets chip companies design RISC-based chips without paying any licensing fees (or royalties).
Just to give an idea of how relevant RISC-V is going to be, we can mention companies like Huawei and Google as first supporters of the project and even the same NVIDIA, which, before deciding to acquire Arm, based their next-generation Falcon logic controller on RISC-V.
Back in 2010, computer science professors Krste Asanović and David Patterson (I remember studying on one of his outstanding textbooks at the university), along with a group of graduate volunteer students at the University of California, Berkeley, decided to develop and publish an open-source compute architecture. What triggered them was the lack of innovation in the RISC-based available architectures and the frustration of seeing CPU vendors charging very high royalties and fees for their use in end products.
In 2015, the non-profit RISC-V Foundation was established and their first official ISA was published so that everyone could download the chip design and freely use it. As already mentioned, companies don't have to pay a fee to use the RISC-V architecture, no matter how many times they download it, and no matter how many devices they decide to put the design in.
Tech companies all over the world are rushing in order to have their first chip based on RISC-V, which would save them tons of money and protect them from NVIDIA's dominance.
Of course the reality is not black and white. Arm's chips still have a big value, and they have already been used in countless devices, included the ones which are currently under development.
Moreover, not all the tech companies (especially pure software ones) have the skills to pick a microprocessor design, implement and integrate it into their chip designs. Arm's expertise still counts a lot for those companies.
So Arm's (or, possibly in the future, NVIDIA's) moat is still there, but the moat is most probably going to be eroded by the rise of RISC-V based chips. Semico Research estimates that by 2025 there will be over 62 billion RISC-V-based devices shipped globally.
The advent of RISC-V is also shaping the way companies behave in the ecosystem, with some of them like Western Digital releasing their RISC-V based SweRV CoreTM to the open-source community and others, like Siemens, offering RISC-V Toolchain services to other companies with less experience in the sector. In computer programming, a Toolchain is a set of programming tools used in development activities (for example a code compiler) aimed at creating a software product on a specific platform.
Conclusions
NVIDIA's planned acquisition of Arm Holdings from Softbank is forming worrying waves in the tech ecosystem as it could result in diminished competition and margins reduction for Arm's customers, with some of them openly opposing the deal.
On the other side, whether the deal goes through or not, Arm Holdings will have to face the increasing appeal and rising adoption of RISC-V based designs in their customers' end products. Arm Holdings is already reacting by lowering their licensing fees to fend off RISC-V competition.
I think in the next few years their currently solid moat will most probably be largely eroded by the adoption of RISC-V (either because of replacement or price pressure), with a consequent shrink in profit margins and market share.
Disclosure: The author does not own shares in any of the mentioned companies
- Genmab: A Company With Downside Antibodies
- How the 'Fear of Mising Out' Can Harm Your Profits
- Graftech: Mitigated Risk and Improved Industry Fundamentals​
Not a Premium Member of GuruFocus? Sign up for a free 7-day trial here.
