Tech Newsround – A Volatile Night

Alphabet Q3 26: GCP almost saves the day.  

• Alphabet reported generally good results but lower monetisation in AI Mode and negative free cash flow for only the second time ever spooked the market, sending the shares down 10% in after-hours trading.
• Q3 26 revenues / operating profit were $127.6bn / $40.6 behind expectations of $128.2bn / $43.2bn as the search advertising business has entered the long-feared transition.
• This was compounded by cash flow jitters, where $42.3bn of cash flow from operations and $37.7bn of debt and equity issuance was eaten by $51.4bn of capex, resulting in no change in cash compared to Q2 26.
• Despite the massive investments, Alphabet’s balance sheet remains in pretty good shape with $139bn in net cash at the end of Q3 26 compared to $81.3bn in Q3 25.
• This is why I think that the debt and equity raises make no sense, as Alphabet has the reserves to make its investments without having to raise debt.
• Debt gives a company tax efficiency as its cost is deducted before corporate tax is calculated, but it comes with the burden of having to be paid regardless of the business situation.
• Given rising interest rates and the market’s uncertainty surrounding AI investments, Alphabet is doing its shareholders a disservice by continuing to raise debt and equity in this environment.
• However, the real problem is search, where growth fell to 11% YoY from 17% YoY in Q2 26.
• Alphabet tried to pass this off with a number of excuses, but the reality is that user activity is moving to AI mode, which Alphabet has not fully figured out how to monetise properly.
• This has happened several times in the past, like when traffic moved from computers to mobile and when advertising became programmatic.
• Hence, I think that Google will figure this out, and it said that it was pretty happy with the experiments that it has conducted so far, but the short-term obsessed market doesn’t care.
• The one thing that went well was GCP, which blew estimates away with growth accelerating to 103% YoY, but this does include sales of its silicon chips (TPUs) to other AI data centre providers.
• This is the AI investments made in 2025 starting to bear fruit, and I suspect that there is more to come.
• However, the market is going to remain sceptical until Google shows a reacceleration of search growth when it figures out the best way to monetise AI Mode.
• Hence, this may be a good point to consider taking a position in Alphabet as the shares are now trading in the mid-teens in terms of 12-month forward PER.

Microsoft FQ1 27 – The one steady report

• Microsoft reported good results, spent less money and did not issue financial instruments, resulting in a good reception by the market that sent the shares up 4% in after-hours trading.
• FQ1 27 revenues / Adj-EPS were $92.3bn / $4.89, nicely ahead of forecasts of $90.6bn / $4.72, as Microsoft met or beat expectations in most of its reporting segments
• Azure revenues grew by 48% YoY to $33.2bn ahead of expectations of $32.3bn, and there is very likely more to come.
• Capex came in 12.5% behind expectations at $45.2bn compared to forecasts of $51.7bn, as its projects suffered from delays, permitting problems and power constraints, all of which mean that it takes longer to roll out new data centres than previously forecast
• Microsoft stated that capex would accelerate again in fiscal H2 27 due to memory pricing, but I suspect that delays are going to continue to hamper the rollout, meaning that cash flow generation will continue to outpace expectations.
• Microsoft also guided strongly for the coming quarter, with Azure accelerating again to 50% YoY, driving FQ2 27 revenues to $97.1bn ahead of forecasts of $95.6bn.
• The net result was a steady quarter from Microsoft, where AI delays helped strengthen the balance sheet.
• In this environment, this is perceived as a good thing, but at 27.8x fiscal 2027 PER, I would rather look elsewhere.

ServiceNow Q3 26 – SaaSpocalypse zombie awakens!

• ServiceNow reported excellent results but a tiny miss in one forward-looking metric re-awakened the SaaSpocalypse zombie, sending the shares down 18% in a move that I think gets reversed  fairly quickly.
• Q3 26 revenues / Adj-EPS were $4.1bn (up 21% YoY) / $1.00, broadly in line with estimates of $4.09bn / $1.02 as AI-related business continued to expand very quickly.
• Q4 26 revenue guidance was also positive, with subscription revenue growth of around 23% YoY forecasted, which gives Q4 26 revenues of around $4.36bn in line with consensus of $4.36bn.
• AI revenue targets have been increased yet again for 2026, with margins and cash flow all heading in the right direction.
• However, current revenue performance obligations (cRPO), a measure of order book strength, grew by 20.5% in Q3 26 but will slow down in Q4 26 to 18.5%, which is 50bp below the consensus estimate of 19%.
• This was enough to awaken the SaaSpocalyspe zombie, which then proceeded to eat 18% of the company’s market cap or $29bn.
• The fear here is slowing growth or revenue declines as clients cancel their contracts with ServiceNow and use AI to run their businesses instead.
• PocketOS has already demonstrated what a foolish thing this is to do, and ServiceNow’s AI revenue growth is telling us that exactly the opposite is going on.
• The actual cause of the miss was an accounting issue where ServiceNow is including delivery expertise with its platform offering to get clients up and running more quickly, and so it is accounted for as a professional service rather than a subscription, causing the tiny miss.
• These services switch back to subscriptions upon renewal, and so there is really nothing to see here.
• Hence, I see this fall in the share price as a great entry point, as the value proposition that supports a valuation of around $233 per share remains intact and may even increase a little bit.
• Hence, I think this anomaly will get reversed reasonably quickly.
• I have a full position in ServiceNow, which is the only reason why I am not buying any more.

Oracle Investor Day – Surviving not thriving.

• Oracle gave some long-term forecasts that reassured the market that things are not as bad as they look, but I still think that Oracle has massively undersold its AI compute to OpenAI, meaning that the real benefits of Stargate accrue to OpenAI, not Oracle.
• Capital expenditure is expected to peak in 2028 and be self-funded from 2029, while free cash flow is expected to increase to $30bn by the end of 2029.
• This allowed the shares to rally 12% and the bonds to increase from $0.79 on the dollar to $0.83, giving comfort that the company is not going bankrupt, but whether it is going to make a decent return on its investments remains to be seen.
• This is because I estimate that the term rate for compute capacity is around $25bn per GW, which ensures a good return, but Oracle has pre-sold its capacity at around $14bn per GW to OpenAI.
• Returns at this price point are pretty meagre on Blackwell and non-existent on Vera Rubin, and given that Oracle is expected to build something in the order of 10GW, $30bn in free cash flow looks pretty weak to me (6% yield).
• This is why I think it will do everything in its power to renegotiate its contracts with OpenAI, which is what the force majeure notice to Blue Owl Capital was partly about.
• Hence, the catalyst for Oracle to really improve its outlook is a contract renegotiation rather than the financial forecasts it has given here.
• These disclosures certainly reduce the probability of Oracle going bankrupt, but surviving is one thing while thriving is quite another.
• I am staying on the sidelines.