Qualcomm FQ4 & Arm FQ2 27– Horrible Handsets

Qualcomm FQ4 26 – Wall of Worry

• Qualcomm reported good FQ4 26 results, and it is obvious to anyone that growth is about to accelerate, although component pricing and handsets are going to be gadflies for another quarter or two.
• FQ4 26 revenues / Adj-EPS were $10.3bn / $2.17, nicely in line with forecasts of $10.1bn / $2.16, but guidance was a tiny fraction short.
• Here, FQ1 27 revenues will be $10.4bn – $11.2bn ($10.8bn) / $2.15 – $2.35 ($2.25) ahead of the consensus revenue forecast of $10.6bn but 3% short of Adj-EPS forecasts of $2.32.
• For this grave sin, the market reduced the valuation of the company by 10%, creating yet another attractive entry point for anyone still sitting on the fence.
• While FQ4 26 handset chip sales were down 13% YoY, automotive put in another mighty quarter with 89% YoY growth, while IoT (bucket for everything else) managed 8% YoY.
• The data centre business also kicks off revenues this quarter with $650m targeted for FQ1 27, and I think that the FY27 target of $5bn is already in the bag.
• Add all this up, and in FY27, the non-handset business will become the majority of chip sales, meaning that going forward wobbles or recoveries in the handset market will affect the company less and less.
• Apple is also likely to vanish completely as a customer this year to be replaced by SpaceX, which, in my opinion, is a much more interesting customer to have.
• This is because Starlink is expanding to offer a mobile phone service, which means that a cellular network is needed to cover all the places where satellite won’t work, like indoors and in vehicles.
• It looks to me like Qualcomm will be the sole supplier of the processor chip, radio chip and the Level 1 software that SpaceX needs to be able to build its own basestation.
• This is not going to trigger a big jump in revenue, but it will be significant in fiscal 2028 and is a major endorsement of Qualcomm’s technology.
• Qualcomm is facing the same problem as Nvidia, which took a share price hit from the weaker gross margin forecast and where no one gave it credit for the recovery to come from the price increases that it put through.
• Qualcomm announced increases some time ago, but these things take time, with the March quarter seeing some benefit to gross margin and the full benefit and a return to 45% – 50% gross margins in the June quarter.
• The market is giving Qualcomm no credit for this, as the shares are trading on 12.7x FY2027 earnings, making it almost as cheap as Nvidia.
• However, I think Qualcomm is under-owned by the market, whereas everyone already owns Nvidia, and so it is much easier to see a wall of money coming at Qualcomm when the market finally changes its mind.
• I can get to $310 per share without trying very hard, and so I remain very happy to sit on this one.

Arm FQ2 27 – Getting interesting

• Arm reported good FQ2 27 results and encouraging developments in the data centre, but the short-term-obsessed market was not very impressed, meaning that Arm’s shares continue to get cheaper as the business expands.
• FQ2 27 revenues / Adj-EPS were $1.44bn (up 27% YoY) / $0.52 (up 33% YoY) as penetration of non-handset market segments continues to make up for weakness in the smartphone market.
• The real driver of growth was licensing, which increased by 50% YoY to $773m, which is a great indicator of what is still to come.
• Most of Arm’s business works with an upfront license when a product decision is taken, followed by royalties as the products start to sell in the market.
• Hence, strong licence growth is a good sign that royalties will start to pick up in the short to medium term.
• Arm reckons that its share in the data centre will pass 60% by the end of calendar 2026 and I see no reason at all why it cannot go higher.
• The data centre moat has been smashed with the move to AI, and Intel has a lot of work to do to catch up, as even its partner Nvidia constantly shows unfavourable comparisons to x86.
• Intel is confident it can catch up, and given progress to date, I am happy to keep an open mind.
• However, in the short term, like Qualcomm, the environment is a little tricky as a large portion of revenues still come from smartphones, which continue to suffer from high memory pricing.
• FQ3 27 revenues / Adj-EPS will be $1.54bn (+/-$50m) / $0.56 (+/- $0.04), which is broadly in line with forecasts of $1.53bn / $0.55, explaining why the shares did very little following the numbers.
• Arm continues to expect that its new AGI CPU designed for data centres will deliver $2bn of revenue in FY 2027 and FY 2028, with the bulk of the revenue coming in FY 2028.
• Competition at the moment is almost irrelevant, as everyone is selling everything that they can make and could sell more if they had the capacity, but this will change over time.
• However, Arm has already signed up most of the hyperscalers, and it is clear that they are all following a diversified approach when it comes to the non-GPU part of the data centre business.
• This is the lynchpin of Arm’s CPU offering, as winning the first clients for a completely new silicon chip is always the most difficult part.
• Arm has already cleared this hurdle, and so the outlook for its CPU business is that it will be able to sell as much as it can make for at least as long as the current shortage lasts.
• As everyone is already sold out for 2027 and is moving into selling 2028, the outlook for Arm’s fiscal 2027 and 2028 (ending March 2028) remains very healthy.
• Hence, it really comes down to a question of upside and valuation, and clearly, as a beneficiary, no matter who supplies the CPU, its earnings should command a premium to the market.  
• Arm is now the cheapest it has ever been, since it relisted at 82.9x FY 2027 PER and 61.1x FY 2028 PER as a result of the decline in the share price as well as the growth in earnings.
• On that basis, I think it is starting to become interesting, but with Qualcomm and Nvidia now back below 13x 2027 PER, the case for investing in Arm over these two is still difficult to make.