Intel Q2 Earnings: CPU Demand and Foundry Execution Await Validation
0xBroomberg
Intel reports Q2 after the bell on July 23, guiding for $14.3 billion revenue at midpoint and $0.20 non-GAAP EPS — but with the stock already up over 160% this year at roughly 87× forward P/E, the market's real bar sits far above guidance.
Up 160% — can guidance hold the line?
Q1 revenue hit $13.6 billion, up 7% year-over-year, beating its own forecast for a sixth straight quarter. Q2 guidance: $13.8 billion–$14.8 billion.
As of July 17 the stock traded near $95, up over 160% year-to-date — while the Philadelphia Semiconductor Index fell more than 20% from its June peak, entering a technical bear market.
This means → Intel rallied against a sector-wide pullback; valuation is running ahead of earnings. Any miss, however slight, amplifies the pullback risk.
HSBC raised its target 100% to $200 — the highest on the Street — which also shows how extreme the analyst disagreement has become.
Why does the CPU suddenly matter again?
For two years the AI narrative belonged almost entirely to GPUs. But as AI-agent workloads — software frameworks that let AI execute multi-step tasks autonomously — scale up, CPUs are re-emerging as a throughput bottleneck.
CEO Lip-Bu Tan told the JPMorgan TMT conference that the CPU-to-GPU ratio is moving from 1:8 toward 1:1, and could reach 4:1.
In plain terms = where one CPU once served eight GPUs, the ratio may flip to four CPUs per GPU — an order-of-magnitude jump in CPU demand.
The data-center-grade Xeon 6980P list price rose to $13,955, up $1,495 from 2025. Intel raised select SKUs, not the full lineup — this reflects a product-mix shift upward under constrained supply.
What do the data-center numbers say?
Q1 data-center and AI revenue reached $5.1 billion, up 22% year-over-year — far outpacing the client-computing segment's 1% growth.
TSMC posted Q2 revenue of $40.2 billion at a 67.7% gross margin; management likewise flagged AI agents as a driver of data-center CPU demand.
This means → Intel's "CPU revival" thesis is not a solo claim. TSMC's industry-level data provides cross-validation — the demand signal is real.
Foundry: the numbers look good — where's the cash?
Q1 foundry-segment revenue was roughly $5.421 billion, up 16% — but about $5.251 billion of that was inter-segment elimination. External customer revenue was negligible.
The segment posted a $2.437 billion operating loss in the same quarter.
In plain terms = the vast majority of foundry revenue is Intel selling to itself. Real outside money is barely trickling in, and the unit is deep in the red.
Apple, Tesla, SpaceX, xAI, and Nvidia are potential foundry customers, but should be treated as long-dated option value, not locked-in orders. The gap between customer test runs and full-scale production can stretch years.
18A: risk production is not mass production
18A-P — the enhanced variant of Intel's next-generation process node, 18A — entered risk production in June, delivering 9% higher performance at the same power, or 18% lower power at the same performance.
Lip-Bu Tan said 18A already supports Panther Lake mass production, with yield — the share of good chips per wafer — improving roughly 7% per month, ahead of internal targets.
This means → the technical metrics are progressing, but a vast gap remains between risk production and high-volume manufacturing. The earnings-call language on yield ramp and volume timelines may move the stock more than the quarter's EPS.
Three scenarios — which way does the stock go?
Bull: Q2 revenue and EPS beat guidance + data-center acceleration continues + gross margin improves + management confirms 18A yield and external-customer progress.
Neutral: CPU business stays strong but foundry losses and production timelines show no clear improvement — the stock stays volatile.
Bear: 18A delays, flat guidance, or margin pressure — in a sector already compressing high multiples, the stock could still fall roughly 30%.
This reflects the real test of this earnings print: not whether AI demand is growing, but whether Intel can convert that demand into deliverable capacity, sustainable profit, and external foundry cash flow — the core proof point for an 87× forward P/E.
Content is for reference only, not financial advice.