SoftBank & Arm – IPO conundrum

One of the trickiest IPOs for many years.  

• SoftBank’s difficulties are increasing after the Chinese economy has hammered the valuation at which it could exit Alibaba and the knock-on impact on the smartphone market has hit Arm’s financial performance at precisely the worst time.
• SoftBank has applied to the NASDAQ to readmit Arm to the public markets in September 2023 at a valuation of $60bn – $70bn which is going to be very demanding when revenues are declining even if it is for a short period of time.
• The culprit here is macroeconomics driven by inflation in the West combined with the failure of the Chinese economy to bounce back from Covid.
• SoftBank calls out these drivers in its FQ1 24 results where the inventory correction in completed silicon chips and weak consumer spending on smartphones were blamed for a 10.8% decline in revenues in USD terms.
• Royalties declined by 19.3% YoY (mostly smartphones) but other revenues actually improved by 3.6% YoY as Arm has been signing new licenses in areas outside of smartphones.
• This is where SoftBank will push the AI story as Nvidia (by far the leader in AI training chips) is launching new products which also use Arm’s IP.
• Furthermore, RFM research indicates that the real action in AI is going to be in inference and not training for models that operate at scale.
• For example, it has been estimated that OpenAI is spending $700,000 per day to support the free version of ChatGPT because it has so many users (see here).
• RFM research has concluded (see here) that generative AI services running at scale will be far cheaper to deliver if the inference is executed on the end device rather than in the cloud.
• This makes a great case for running inference on smartphones (which Qualcomm and others are already working on) and not in the cloud.
• Given that Arm is present in almost every smartphone in existence this puts it in a very good position to have its technology adopted for generative AI inference on battery-powered devices.
• However, in terms of time, this remains some time away and when one looks at the proportion of Arm’s revenues that are coming from AI and generative AI today, they are pretty small.
• Arm has the same problem as Qualcomm in that the long-term looks good, but the smartphone market is playing merry hell with its immediate-term performance.
• The other issue is that the market is notoriously fickle when it comes to stocks which demand high valuation multiples, and a tiny wobble can cause the valuation to quickly unravel.
• In fiscal 2021 Arms USD revenues grew by 35% to USD2.7bn and in 2022 by 6% to $2.8bn but in 2023 they may well decline unless there is a sudden reversal in the smartphone market.
• Qualcomm stated on its conference call for its FQ3 23 results that saw no immediate sign of recovery and that the inventory correction may last until the end of the calendar year.
• This does not bode well for Arm returning to growth this side of 2024 which is going to raise questions from investors being asked to pay 23x FY 2022 revenues for a business where revenues are declining given how short-term focused the market is.
• This is why the AI story is so important to getting this IPO away at the desired valuation and explains why SoftBank is looking for anchor investors (who will pay more (see here)) and also could be behind its potential purchase of the 25% of Arm that it doesn’t own from Vision Fund 1.
• SoftBank is doing everything it can to maximise the valuation of Arm at IPO, but the rubber will not really meet the road until it becomes clear what valuation the market is willing to bear.
• If the AI story sticks, then this will probably get away comfortably given the ludicrously high valuations that the market will afford to those seen to be leaders in the generative AI race.
• If not, then SoftBank has a very difficult task on its hands and may end up having to settle for much less.