Server CPU demand is outpacing GPU orders for four straight record quarters — and a looming ABF substrate shortage by 2027 could become the physical ceiling on AI deployment. Today's briefing covers AMD's earnings signal, Arm's $20B AGI demand constraint, Intel's foundry repricing, and the packaging bottleneck hiding in plain sight.
Audio is available on Spreaker — see link below.
AMD just confirmed what the GPU-obsessed market has been slow to price in. Server CPU demand is growing faster than accelerator demand.
Arm posted record quarterly revenue of one point four nine billion dollars and CEO Rene Haas put a number on something the market had been treating as vague upside. AGI CPU demand exceeds twenty billion dollars.
Intel jumped fifteen percent in a single session on reports of a preliminary chip manufacturing agreement with Apple. That pushed its one-month gain to one hundred and thirteen percent.
Underneath all of this sits a packaging problem that's moving from logistics footnote to board-level risk. ABF substrates, the material that connects chips to circuit boards in high-performance packages, are on track to fall short of demand by 2027 as AI chip volumes grow at sixteen percent compound annual growth rate.
The architectural question underneath all of this is genuinely unresolved. Intel Xeon and AMD EPYC benefit from decades of software certification.
One number that deserves more attention than it's getting: Cisco posted two point one billion dollars in hyperscaler AI infrastructure orders in its most recent quarter, with networking revenue up twenty-one percent year on year. The GPU vendors absorb most of the analyst coverage, but the networking layer is where AI cluster costs compound.
The CPU narrative has crystallized this cycle. What was speculative is now confirmed by earnings.
Chapter summary auto-generated from the verified script. Listen to the full episode for the complete content.