Meta & Microsoft – Spend, Spend, Spend.  

Microsoft FQ2 26 – Wobble at the top.

• Microsoft reported good results but spooked the market with a huge capex bill while at the same time raising concerns about return on investment, given that Azure did not blow the doors off.
• FQ2 26 revenues / Adj-EPS were $81.2bn / $4.14, comfortably ahead of estimates at $80.3bn / $3.91.
• The business continues to fire on all cylinders with Azure and cloud once again taking the front row seat.
• Azure saw 39% YoY growth, which was just in line with expectations but a little bit shy of what the market was hoping for.
• This is because capex remains high at $37.5bn in FQ2 26 alone, and there was some hope that there would be some acceleration of growth as a result.
• This did not occur, and Microsoft explained this as investments in AI capacity also being used for internal purposes, such as bolstering other products, which sounds like a fairly lame excuse to me.
• I suspect that the reality is that capex projects always seem to take longer than expected, and the increase in growth that was expected will occur, but just a bit later than the impatient market was hoping for.
• To be fair to Microsoft, it badly needs to upgrade its internal products, as the search function in Outlook remains pathetically bad, Office Help is as useless as ever, and CoPilot remains an also-ran with not much usage.
• This won’t have much effect in the short-term, but Google will be looking to see how it can make its productivity offering better with AI, and Microsoft will need to be at least as good just to hold its current position.
• All eyes now turn to Google and Amazon as this will tell us how the overall cloud services market is developing, and I expect positive news from Google and neutral to negative news from Amazon.
• Microsoft has lost 2^nd place in the valuation table to Google, which now trades at around 33x 2026 PER while Microsoft has fallen meaningfully to around 28x 2026 PER.
• If this continues, I may look to get back into Microsoft, but at the moment, the one to really look at is Meta, which has gone from the value play to outrageously expensive (see below).

Meta Q4 25 – Peak Splurge.

• Another excellent quarter with the core businesses firing on all cylinders, disguising the blow of further massive capital expenditure, which will cause earnings to crater, meaning that the 2026 consensus EPS estimate of $29.66 is way too high.
• Q4 26 revenues / EPS were $59.9bn / $8.88, nicely ahead of estimates of $58.5bn / $8.21.
• Meta now has 3.5bn DaUs within its family of apps, which, combined with improved advertising targeting, allowed revenues to grow by 25% YoY.
• What was unusual was that the price per advertisement actually increased by 6% YoY, which is unusual in this industry and is a sign of where the heavy investments in AI are going.
• This is what Meta Compute is all about, which was announced earlier this month.
• This is a new division within Meta that is tasked with providing 10s of gigawatts of compute by 2030 and hundreds of gigawatts over the coming decades.
• A key focus of this effort is vertical integration, where Meta is working on both in-house silicon as well as accessing independent sources of power, such as small modular nuclear reactors.
• This strongly implies that there is going to be a focus outside of the Nvidia ecosystem where Broadcom, MediaTek and Qualcomm are all jockeying for position.
• With 10s of gigawatts up for grabs over the next 5 years, the rewards could be very large indeed, even if there is a correction and AI demand gets delayed by several years.
• In the meantime, the better-than-expected profitability has boosted cash flow from operations, which at $36.2bn is enough to pay for the massive capex bill.
• However, Mr Zuckerberg is doubling down on his commitment in announcing that expenses in 2026 would amount to a gargantuan $162bn – $169bn most of which is driven by 3^rd party data centre costs, depreciation on capital investments and higher operating costs of new data centres.
• With a revenue estimate of $221bn for 2026, gross margins of 85%, and expenses of $165.5, this means that EBIT will be $22.4bn down 73% compared to 2025.
• This means that 2026 net income will be around $19.5 bn or $7.61 per share, down some 67% compared to 2025.
• This looks set to continue for the foreseeable future, meaning that a very large acceleration in revenues is required to offset this huge increase in expenses and of this, there is no immediate sign.
• This puts the shares on 93.7x 2026 and probably something similar for 2027 and 2028.
• Hence, I am moving from indifferent to negative on the shares and would use the bounce to sell the shares and go elsewhere.
• The beneficiaries of the latest spending spurge will be the usual suspects, such as Nvidia, AMD, Broadcom, MediaTek, Qualcomm, Micron, SK Hynix and Samsung, and I would immediately switch into these as even after stellar runs, they all look like better value.
• I continue to hold both Qualcomm and Samsung in my portfolio.