It should quickly become official: Nvidia should throw in the towel and abandon its offer to buy ARM for 40 billion dollars. Announced in the summer of 2020, the deal raised many eyebrows. Both that of one of the founders of ARM and that of almost all the competition authorities.
A semi-failure for Nvidia – which has seen its stock market quotation more than double since the announcement – as well as for Softbank, which now has a jewel worth, at the very least, 40 billion dollars. Recall that he bought the company for “only” $32.4 billion in 2016.
Read also: From smartphones to supercomputers, can ARM processors dominate the world?
Lobbyists for giants like Apple, Google, Qualcomm, Broadcomm, and even Intel have no doubt worked tirelessly to derail a case that was only one for Nvidia.
With more than 500 customers integrating its processor architectures into smartphones, supercomputers and even refrigerators, ARM could not be bought out and controlled by one of the most powerful players in the world of semiconductors. Especially when you know the natural aggressiveness of the flagship GPU that is Nvidia.
So the deal fell apart, everyone goes home, and nothing has changed? Nay. If on the one hand we still have to wait for confirmation of this abandonment from the mouth of Nvidia itself – we told you, Nvidia is a very voluntary company – and if it is obvious that Softbank will proceed with an introduction on the stock market or find a more “neutral” buyer, the situation has however changed a lot in 18 months of “soap opera”.
ARM is still for sale
We forget too often: if Nvidia put 40 billion on the table, it was because ARM was for sale. More precisely it is because Softbank, its owner who had bought ARM in 2016, wanted to do somersault and continue to invest in other, even more lucrative segments.
We must therefore already count on the financiers of the Japanese firm to study their plans. The IPO is the most leaked scenario because on paper it is the most viable project, now even more so after such a high bid. Short of a fund or a bank, which sufficiently neutral actor has 40 billion to invest in intellectual properties around processor architectures?
A failure that validates the importance of ARM
When the takeover offer was made public in the summer of 2020, it shocked the tech world. Gradually, tech companies, but also journalists, analysts, specialists, as well as companies that depend on tech (automotive in particular), and many public authorities have measured the importance of ARM. And, even more, the importance of an independent ARM.
Read also: Why Qualcomm hasn’t (yet) announced ARM chips that compete with Apple’s M1s
A company that has now seen its enormous financial and strategic value validated. If this limits the contenders for a complete takeover, it does not in any way hinder an IPO of the title at its “fair price”: that of the most powerful ISA (the instruction sets) on earth. But, because there is a but, this power is now well integrated and does not only have advantages…
The weakness of the ARM model highlighted
When we depend too much on Intel in the x86, we can always go to see AMD, which also does very good work in this type of chip. If the duopoly is not perfect, at least there is increased competition in recent years.
On paper, the ARM model is even better since if you don’t like Qualcomm you can go to MediaTek, Samsung, Rockchip or (formerly) HiSilicon.
If in fact the equation is more complicated than that, it is the idea of the model to be able to change supplier of ARM chips. And, for some companies, to be able to count on an ISA that you can “tinker” at your leisure (by paying for the right license) to make your chips with little onions.
Except that the Huawei Gate affair was an opportunity for Trump’s America to deprive the Chinese champion of ARM. An attempt that did not work – in the end, it was at the level of the factories that the United States struck – but which has already provoked many reactions as to the independence of a British company in the face of American pressure.
True technology independence may be called RISC-V
If there’s one company that loves its independence when it comes to technologies, suppliers, and even factories, it’s Apple. In the Nvidia-ARM adventure, Apple obviously measured how much of its technological independence could have been called into question.
Of course, Nvidia had sworn-spit that it would not touch the business model of ARM, and warranted that it would maintain professional relationships with its customers. However, we could legitimately doubt it in the long term.
Apple, more than others. The tech titan has completely cut ties with Nvidia following a dispute over the heating of certain GPUs in MacBooks. If it is impossible to know what the plans of the very secretive giant are, one does not have to be very sly to see what the “miracle” solution would be: the ISA RISC-V.
Read also: Why Apple is hiring a RISC-V engineer, and why it matters
Originally developed by an American university, this processor “architecture” – in reality, once again an ISA, that is to say an organization and instruction sets – has become open. Little by little, in recent years, it has become a foundation that has migrated from the United States to Switzerland to guarantee its freedom, openness and neutrality (Huawei-gate effect).
If the ARM ecosystem, with billions of chips, several decades of evolutions, developments, software tools, teaching programs in universities, etc. is infinitely superior in power and resources, RISC-V continues to grow in power.
ISA is cited and studied in space research programs (in succession to SPARC), players like SiFive are getting so big they are valued in the billions, and many commercial products now ship not with ARM cores, but with RISC-V cores.
ARM took note of this rise in power and even fomented a communication campaign denigrating RISC-V.
Also see video:
Also see video:
Because, if we are more than ever dependent on the power of ARM, it is perhaps precisely because of this power that the tech world will bring out a real alternative. And if that happens in the next five years, we can say thank you to Nvidia for having accelerated the dynamic…
Read also: the Nvidia / ARM soap opera