Will Nvidia’s $40 billion deal with Arm leave processor licensees wringing their hands?
You might say that the signing of a definitive agreement for Nvidia to acquire Arm for $40B represents ‘the chickens coming home to roost’. Few would argue that the processor at the heart of 160 billion electronic devices is not a strategic component of the system. Yet, contrary to established business wisdom (that businesses should retain control of their strategic assets), the chip industry has been content to divest itself of processing technology. Kudos to Arm, who have crafted a proposition attractive enough to seduce even the biggest chip companies into following this path. Would they have made the same decision if the processor IP was being licensed by a competitor, say Nvidia, for example?
As far as I can see, this seismic announcement opens three potential paths to electronic manufacturers: 1) keep to business as usual, 2) pivot to another licensed architecture, or 3) gain back control of this most important of assets. Sure, Jensen Huang has promised to retain an “open-licensing model and customer neutrality”, but can the industry afford to continue to put its faith (and fortune) in this assurance? Does he mean that Nvidia will remain as one of Arm’s customers, keeping a level playing field for all? Business-as-usual seems an unlikely strategy for the largest Arm licensees, for whom this news will be a massive wake-up call.
Perhaps some will pivot to other licensed cores. RISC-V is immature compared to Arm, but benefits from an open and free instruction-set. Ironically, its biggest benefit may hamper its growth – RISC-V is already fragmented and may take many years to mature, to the point we see well supported configurations really start to emerge (think Linux and Redhat). In the meantime, it is unlikely to benefit from the collective attention that would result from a wholesale pivot of the chip business.
Other licensing businesses may also have an answer – ARC processors from Synopsys, Tensilica processors from Cadence – these are Electronic Design and Automation (EDA) companies that used to be part of a “big-3” with Mentor. Where is Mentor? Acquired by Siemens in 2017. In the last few years many deals have demonstrated that there is no company so large that it cannot be acquired. Licensing another processor from another company may just be jumping out of the frying pan and into the fire.
The last option, and surely the most strategically robust, is for chip companies to retake ownership of their processor architectures. The cost of doing so is significant, but when measured against the existential threat of lost or compromised access to such an important part of their products, it seems like the only real and safe choice.
Mark Lippett, CEO, XMOS