Intel & AMD – Credibility Stretch

Intel and AMD both require a stretch of the imagination

Intel Q2 26 – Breathing space.

• Intel is riding the CPU wave as a virtue of being in the right place at the right time, and while it still has serious problems, it should be able to generate cash to invest in the turnaround.
• Q2 26 revenue / Adj-EPS were $16.1bn (up 25% YoY) / $0.42, comfortably above expectations of ($14.4bn / $0.22) as demand for CPUs continues to expand as the AI market moves away from training and towards inference for agents.
• This strength is going to continue, which allowed Intel to increase its guidance for the coming three months, where Q3 26 revenues / Adj-EPS are expected to be $15.8bn – $16.8bn / $0.38, which is nicely ahead of expectations of $15.6bn / $0.32.
• The good news in this report is the cash flow statement, where cash flow from operations improved to $8bn, which gives Intel badly needed breathing space to right the ship.
• The problem here remains unchanged in that Intel’s core x86 technology increasingly looks obsolete, with Intel’s partners constantly comparing their Arm designs very favourably against x86.
• This remains the key problem to solve, where the good news is that Intel is generating cash that it can use to solve the problem, but the bad news is that it is showing no indication of knowing how it will solve the problem.
• For the moment, the market does not seem to care, as all it sees is x86 CPUs selling again and is not concerned that at some point these designs may run out of road.
• This is not a bet I am willing to make, so I am staying away from this one.

AMD – The Big Leagues.

• AMD has stepped up its challenge to Nvidia with the release of new products, a partnership with Cerebras as well as market guidance that is pretty much in line with what Arm and Nvidia have already hinted at.
• AMD put some hard numbers on the markets that it serves, with the AI accelerator market expected to reach $1.4tn in 2030, with CPUs for AI reaching $220bn.
• These are pretty wild figures because if one assumes that silicon remains 75% of the capex spend on data centres, this means that data centre operators are expected to spend $2.16tn on capex in 2030 alone.
• This is 3x the level that is being spent in 2026, which is already stretching the ability of the most profitable and successful companies in history to pay for it.
• Hence, how they will be able to triple capex from here remains something of a mystery to me and leads me to look at this number with an increasing degree of scepticism.
• At the same time, AMD also launched a host of new products aimed at competing with Nvidia’s Vera Rubin, but it still remains behind when it comes to time, as Nvidia launched Vera Rubin over 15 months ago with products starting to come on stream now.
• AMD also launched a partnership with Cerebras to offer a product line aimed at those who want fast inference, which brings it into line with what Nvidia is doing with Groq.
• It has also announced another one of those “circular” deals where Anthropic will buy 2GW of AI capacity from AMD and in return AMD will become an investor in Anthropic.
• I don’t think that these deals are as circular as people think, as all it really amounts to is Anthropic paying for the chips with shares rather than cash.
• It does create interdependencies though, meaning that any shockwaves will affect everybody and could trigger an industry-wide collapse if the shockwave is large enough.
• The net result is that with its new products and the customers it has lined up, AMD, its chips and the software platform it offers are all increasingly credible alternatives to Nvidia.
• Hence, I expect to see much greater developer adoption of AMD’s ecosystem and the CUDA moat becoming shallower.
• This has been on the cards for a while, especially as the market has moved to inference away from training, where CUDA is less sticky.
• Hence, I don’t see this as a meaningful change in the competitive landscape, and if AMD is able to defy my scepticism, the market is going to be so large that there is plenty of space for everyone.
• I would still prefer Nvidia over AMD due to its market leadership, faster time to market and much cheaper valuation.