What can you do with a spare Arm?

The moment the US Federal Trade Commission (FTC) got involved, the deal was almost certainly doomed. Nvidia has made it official by backing out of the agreement to buy Arm from SoftBank Group. 

Though it will cost $1.25bn to sever the agreement, it is a lot less than the near $70bn and mountains of legal bills the maker of graphics processing units (GPUs) would have to pay to continue to go through with the deal, if it would be allowed at all. Practically every major market in the world was acting on complaints from Nvidia’s competitors. Given that Nvidia’s stewardship might have had customers running, like Intel, into the arms of the open-source instruction set architecture RISC-V, not owning Arm probably is not that big a problem for CEO Jen-Hsun Huang’s company, which itself is a RISC-V user. 

There was some advantage to be had in making Nvidia a broad-based supplier of processors that cover every application. But there is little to stop it doing that now, and it might not even be the right target. Intel’s own embrace of RISC-V may have come too late for that company.

The shift we are seeing now in computer design is less about what processor you pick but how the processor, I/O and memory work together. The specific brand of instruction set is not all that important just as long as the combination works for your application. Techniques like managed code, where many of the instructions are generated or compiled on the fly, make it possible to move from one architecture to architecture far more easily than was the case in the past. Apple, for example, has obtained a lot of experience in this area, having jumped ship from 68K to PowerPC to x86 to Arm over the course of the past 25 years.

This shift leaves SoftBank and Arm with a big problem. Arm was SoftBank's supposedly major strategic purchase of 2016, one that would underpin the conglomerate's myriad technology investments in robotics and the IoT. Chairman and CEO Masayoshi Son may have had diamond hands when it came to Arm at one stage but, because a number of SoftBank's investments have not gone well, the IP supplier has turned into an asset of uncertain value that needs to be flogged pronto. 

The immediate plan is to put Arm back on the stock markets through an initial public offering (IPO). The precise venue or indeed the price is far from certain, but I’m going to bet on the New York Stock Exchange (NYSE) for a lot less than Nvidia’s agreed price – even the one they agreed before the GPU maker’s share price headed for the moon on the back of shareholder enthusiasm for all things artificial intelligence. 

There is no guarantee the planned IPO will even complete, though it will not face the regulatory hurdles that the Nvidia deal did. It is hard to imagine any customers kicking up a fuss about it, but timing is not on SoftBank’s side. It wants to complete the IPO relatively quickly, by the end of March 2023, but it is doing so at a time when semiconductor stocks may be on their way down for a while. The shortages that have helped push stock prices high for the chipmakers are beginning to abate, which will act as a dampener on any IPO offer price. And slowdowns after shortages in this market have a nasty habit of turning into slumps. The likely slowdown in semiconductor sales may have been a factor in Nvidia killing the acquisition now.

Added to that is the effect of monetary tightening on tech stocks in general, which have benefited massively from the low-interest-rate environment of quantitative easing. In an inflationary environment, bets on tech look less tempting. And in an inflationary environment, the discretionary spending that drives much of the electronics sector now comes under much greater pressure.

SoftBank may be faced with the choice of accepting a low offer price or cancelling the IPO and trying to find another private buyer. But the only realistic option there would be private-equity companies who have the option of a more relaxed IPO schedule but who will not pay more than they absolutely have to, especially when regulatory issues are taken into account.

There is another reason why the IPO may turn out to be just as big a headache for SoftBank as the Nvidia deal: China. 

Not long after the Japanese conglomerate took Arm private, the decision was made to set up a joint venture in mainland China, a place where joint ventures are rarely weighted in favour of the international parent. The joint venture’s chairman and CEO Allen Wu then decided to run things his way to such an extent that Arm fired him. Or at least tried to. Wu is still in place at a renamed Arm China that now has a parallel, competing roadmap to the parent company. The situation is far from being resolved, and even with the decoupling from the western economies that China is now going through, it leaves the IP supplier with a major headache over what to do about such a large potential market, albeit one with a patchy record in IP rights.

When Nvidia was the buyer, Arm China was able to fade into the background. It was clearly a problem. But it was going to be Nvidia's problem ultimately. Now, Arm China is a problem that the financial markets will likely see as needing an urgent solution.

The final issue is that Simon Segars, the Arm CEO who has been in place since Warren East departed in mid-2013, is stepping aside in favour of the head of the IP products group Rene Haas. Segars’ replacement as CEO now has the job of stabilising Arm in the wake of a programme of expansion that SoftBank encouraged, dealing with the China issue and making the IP supplier more palatable to institutional buyers on the IPO market.

You wouldn’t really want to start from here.