Qualcomm, Arm, Meta and Microsoft – Results Rush

Qualcomm FQ3 26 – Legacy Issue.

• While the diversification strategy is going very nicely, the weakness in the smartphone market is hurting performance, upon which the fickle market remains laser-focused.
• FQ3 26 revenues / Adj-EPS were $9.95bn / $2.21, broadly in line with forecasts of $9.7bn / $2.23 as memory and supply issues continue to cause structural weakness in the handset market.
• However, the rest of the business continues to go extremely well, with automotive growing 61% YoY and the IoT business (everything else) growing by 9% YoY.
• Guidance was also reasonable, with Q4 26 revenues / adj-EPS expected at $9.7bn – $10.5bn / $2.05 – $2.25, slightly ahead of consensus at $10.1bn, but missing the EPS estimate of $2.34 by 9%.
• This has been caused by the cost pressure in memory as well as the general tight supply environment, but Qualcomm has moved to address this.
• Last week, the company sent letters to its customers stating that there would be a double-digit increase in chip prices from September 1^st 2026, which should go a long way to bringing profitability back within the expected range.
• This was a frustrating quarter for the company, which has just substantially increased its long-term targets thanks to winning 4 hyperscalers for its new datacentre product as well as continued strength in the automotive and industrial businesses.
• AI-related weakness has driven the shares down substantially, representing an opportunity to open a position or to add to an existing one.

Meta FQ2 26 – Zuck stares down the gift horse.

• Meta reported reasonable Q2 26 results, but its indecision on whether to rent out its compute capacity could cost it dearly as the crazy returns that are currently available are not going to last for very long.
• Q2 26 revenues / EPS were $60.8bn / $6.18, missing estimates of $60.2bn / $7.19 as one-off expenses such as severance for the 8,000 layoffs and legal proceedings increased expenses by $3.6bn.
• The combination of higher expenses and massive capital expenditure reduced net cash flow to $784m, meaning that the investment case is becoming a bet on whether Meta will deliver on its AI promises.
• The issue here is that, unlike Google, Microsoft, Amazon and SpaceX, Meta does not generate any revenue from selling compute capacity to others, meaning that its returns from AI will need to be realised through increased revenues and profits from its digital ecosystem.
• This is where I am hoping that Meta will quickly change its mind and sell all of the capacity that it can because the returns are currently so good that it can make a lot of money even if it only sells the capacity for a short time.
• Mr Zuckerberg is considering this opportunity and said on the investor call that he had received plenty of offers to rent his capacity at a substantial premium to what he paid for it.
• Given where the market for compute currently is, it would make sense for Mr Zuckerberg to park his AI efforts and sell the capacity until such time that the price normalises and it makes sense to start working on his own AI once again.
• This large increase in spending without an immediate return is damaging sentiment, and the shares are down 8.6% in pre-trading, which will put some pressure on Mr Zuckerberg to take actions that will please investors.
• Without selling compute, I can get to $700 per share on Meta, with an extra $70 per share should Meta sell up to 5GW of capacity for a 5-year period.
• If it were to do it permanently, this would add $300 per share, but I think that this is probably too high given the inevitable fall in compute pricing that is going to come as more capacity comes online.
• I am holding onto my Meta position as Mr Zuckerberg has seen sense before and I think that he will do so again.

Arm FQ1 27 – Same Legacy Issue.

• Arm reported good FQ1 27 results but did not guide as strongly as some of the street was expecting, as the smartphone market is causing Arm precisely the same problem as Qualcomm.
• FQ1 27 revenues / Adj-EPS were $1.29bn / $0.45, just ahead of expectations of $1.26bn / $0.40, but the smartphone market weakness is causing problems.
• This was apparent in the guidance where FQ2 27 revenues / adj-EPS are expected to be $1.33bn – $1.43bn ($1.38bn) / $0.43 – $0.51 ($0.47), broadly in line with consensus of $1.36bn / $0.46.
• Royalties have been growing around 20% in the smartphone market, but this will now come down to something like 13% to 15% as Arm’s growth drivers are offset by overall volume weakness in the market.
• This overshadowed further good news in the data centre where Arm’s initial shipments of the AGI CPU have been well received, and orders have continued to strengthen.
• Revenues of $1bn that were expected to occur in fiscal 2027 and fiscal 2028 will now be $2bn as both the number of orders and the acceleration of timelines have triggered the uplift.
• Although I remain sceptical on the 2030 expectations as this requires the hyperscalers to spend something in the order of $2.1tn in a single year on data centres, this 21-month forecast looks achievable to me.
• Hence, if the market continues to ditch the AI trade, an opportunity in Arm (that I missed at $100) may once again present itself.

Microsoft FQ4 26 – Sleight of Hand.

• Microsoft reported good results and guided lower on capex thanks to a quirk of accounting that makes absolutely no difference to cash flow, but the market fell for it anyway.  
• FQ4 26 revenues / Adj-EPS were $90bn / $4.81 above estimates of $88bn / $4.25, but this included a $0.33 gain from writing up its investment in Anthropic.
• Even without this, this was still a good beat, which was largely driven by a reacceleration of Azure revenues coming from AI.
• Azure grew revenue by 43% YoY and exceeded $100bn for fiscal 2026, representing a significant closing of the gap to market leader AWS.
• Demand for cloud continues to exceed Microsoft’s ability to supply it, and growth is expected to accelerate again in FQ1 27 to 45% YoY.
• Guidance was broadly in line with expectations, but it was the commentary around capex which pleased the market.
• When it comes to cash out the door, there is no change, but Microsoft now thinks that some of its assets (buildings) will last longer than previously thought.
• This allows it to reclassify some capex as opex and expense it in the income statement instead of capitalising the expense and depreciating it later.
• This will have no effect whatsoever on free cash flow as cash from operations will go down by the same amount that cash from financing goes up (lower capex).
• However, it allows Microsoft to cut the headline capex guidance in an accounting trick for which the fickle and shallow-minded market fell hook, line and sinker.  
• This has been interpreted as Microsoft being able to do more with less, and the shares have rallied by 8% in after-hours trading.
• I still remain concerned with Microsoft’s halting efforts in AI, but it is showing signs of being the harness and the orchestrator for everyone else, which ServiceNow and CRM are also doing and getting good traction.
• I have the same concern with Meta, but Meta is much cheaper than Microsoft, which is on 22x 2027 PER compared to Meta on 15.2x 2027 PER, and Meta also has the potential for upside from selling compute for the first time.
• Hence, I remain pretty ambivalent towards Microsoft and am keeping my eggs in the Meta basket.