Amazon and Intel Q3 26 – Another Bad Night

Results so far are not inspiring the AI trade

Amazon Q3 26 – A cannonball to cash flow

• Amazon is focusing on becoming a cloud infrastructure business as growth has accelerated, but this requires heavier investments, which is hammering cash flow, which is precisely what the market doesn’t like at the moment.
• Q3 26 revenues / EPS were $201.9bn / $1.89, broadly in line with expectations of $202.2bn / $1.97.
• The real story of the results was AWS, where growth has accelerated back to 39% YoY, which, according to Amazon, is where it is going to stay for at least the next two years.
• This was an extraordinary disclosure from Amazon, which usually guides only one quarter in advance, and represents a statement of intent as well as a reassurance to investors that the hundreds of billions that it is investing will generate attractive returns.
• This is underpinned by the backlog of business that AWS has, which has doubled in the last year and now stands at $578bn, which has given AWS unusually good visibility for the medium term.
• However, investors were not reassured as the retail business has seen a significant slowdown, which dragged the overall growth to a much slower 12% YoY.
• The real problem was capex of $77.3bn in a single quarter, which has pushed cash flow deeply into the red despite a $48bn debt raise.
• Capex for 2026 will be $220bn and will rise materially again in 2027, which is going to blow an even bigger hole in the cash flow and balance sheet, as 2027 cash flow from AWS will be $84bn in the best case.
• This means that there are more capital raises to come, and in this environment of rising interest rates, the market is not comfortable with this at all.
• This is the main reason why the shares fell by 7% in after-hours trading even though guidance for Q4 26 was comfortably in line with market expectations.
• The more this goes on, the more one is getting confirmation that 2027 is going to be an even better year for the semiconductor sector than the already legendary 2026.
• Hence, this quarter should be spent in contemplation of what semis to hold in 2027 and using current jitters to lock in good valuations.
• Nvidia is at the top of this list, but also Arm, which has halved since July, GlobalFoundries and maybe one or two neo-clouds which are getting towards interesting territory.
• The picks and shovels trade is still very much the way to go.

Intel Q3 26 – Double Whammy

• Weak PC shipments and supply chain problems have caused shortfalls at both of Intel’s businesses, meaning that expectations for Q4 26 are now well short of expectations, triggering a 23% decline in Intel’s share price in after-hours trading.
• Q3 26 revenues / EPS were $16.5bn / $0.52, broadly in line with expectations for revenue, but EPS was ahead.
• This was driven by better-than-expected performance from the data centre business, where revenues grew by 78% YoY and profitability improved materially.
• The problem is that Intel is in a period of recovery and turnaround, meaning that its grip on its supply chain is not as good as it should be, and it has had problems with packaging substrates, a factory fire in Israel, and wafer sourcing.
• The net result was Q4 26 guidance that materially missed expectations as well as the promise of investments in 2027 eating up all of Intel’s cash flow.
• Q4 26 revenues / Adj-EPS will be $15bn – $16bn / $0.25, well adrift of expectations of $17.2bn / $0.44 and I suspect that this will also cause 2027 profits and cash flow forecasts to be cut.
• However, Intel is right at the sweet spot of the data centre market as models that “think” for longer and use agents require far more CPUs than the regular LLMs that people have been using for several years.
• This is why Intel, Arm, Broadcom and everyone else who makes CPUs sees the market increasing by as much as 5x over the next 5 years or so.
• However, the PC market is really struggling for the exact same reason that data centre demand is exploding.
• Here Intel is really feeling the heat as customers are resizing chip inventories as customers keep their existing computers for longer given how much prices have risen thanks to the memory shortage.
• This is going to persist into 2027, with the only silver lining here being that this gives Intel time to catch up with Arm (as it says it will do in another generation or two) before attention once again focuses on this issue.
• Mr Tan has far exceeded my expectations in terms of the quality of people he has been able to attract to help him effect the turnaround, but he is not out of the woods yet.
• This, combined with a high valuation, uncertainty around x86 longevity and a difficult 2027 still ahead, leaves me cautious on the shares.
• However, the harder it sells off, the more interesting it becomes as my confidence in Mr Tan slowly lifts off rock bottom.