Intel's 25% Revenue Surge, TSMC Price Shock & the Sub-$500 PC Collapse

Jamie Cole (0:00)

A Hardware & Chips, Daily News.

I'm Jamie Cole, thanks for joining me.

Today, memory triage, why AI's HBM hunger is pricing out PC buyers until 2028

Intel just raised CPU prices for the third time in a single year. The latest hike, 10% on list prices, brings the cumulative increase to around 30% year on year. And the cause isn't a supply disruption, a natural disaster, or a demand spike in personal computing. It's AI datacenters eating the memory market alive. Here's the structural reality. High bandwidth memory, the kind stacked inside Nvidia's H-series GPUs, consumes roughly three times the wafer capacity of standard laptop DRM. Memory suppliers aren't blind to that trade-off, they're optimizing for it. HBM margins are significantly higher, so production lines are shifting accordingly. Lpddr5X, the memory that goes into your notebook, surged 89% in a single quarter, as manufacturers chased AI infrastructure demand. The shortage in PC memory isn't incidental, it's the rational outcome of deliberate allocation decisions by suppliers, who have better options.

The important distinction is that this isn't a broken market, it's a market working exactly as designed, just not in the consumer's favor. Gartner's projection makes the stakes concrete, sub-$500 PCs are expected to disappear from the market by 2028 Memory and CPU costs now consume 58% of a notebook's bill of materials.

At that ratio, budget-tier hardware becomes structurally unviable. Meaningful supply relief isn't expected until Micron's US capacity comes online, and that timeline points to 2028 at the earliest. That matters because the entry-level PC market isn't a niche. It's where hundreds of millions of first-time buyers, students and cost-sensitive enterprises sit. Pricing them out isn't a rounding error. It's a segment-level collapse with real consequences for education infrastructure, workforce access and emerging market adoption. Intel's second quarter 2026 results tell you something about how sustainable this strategy looks from the inside. Revenue came in at $16.1 billion, up 25% year-on-year.

Operating profits swung from a $503 million loss to a $2.77 billion gain. That's not product innovation doing the work. That's pricing power. Here's the thing. The signal here is that Intel's financial recovery is running ahead of its competitive recovery. The company still trails on process node differentiation against TSMC-backed rivals. But if memory inflation continues to insulate margin, Intel can report strong numbers without closing that gap. Whether the market rewards that distinction is the unresolved question. Shift to Japan. Rapidus, the $15 billion state-backed chipmaker, announced 17 design partnerships this week. The list includes Synopsys and Infosys alongside 15 others. The target is 2nm production starting in the second half of fiscal 2027 The partnerships matter less as a technical signal and more as a commercial one. Where Pigus is trying to demonstrate customer demand before the fab is fully operational, that's a rational sequencing move. The harder problem is that experts are clear-eyed about the difficulty ahead. Scaling a greenfield fab to profitable volume production is one of the hardest things in manufacturing. A Japanese government official acknowledged that outcome clarity could be 10 years away. 17 design partners is progress. It isn't validation.

Back in the US., the domestic advanced packaging market is starting to show real numbers. A fresh forecast puts it at $11.7 billion by 2031, up from $9.6 billion in 2026 Amkor's planned $7 billion Arizona campus and Intel's FAD9 are anchoring that buildout, with Micron's Idaho expansion adding to the picture.

Consider this, the Chips Act logic is straightforward. Reduce the cost of domestic capital deployment, shift back-end manufacturing onshore, and reduce exposure to Asian supply chain concentration.

That part is working. The remaining vulnerability is softer.

US facilities still depend heavily on Asian suppliers for underfill and dielectric materials. Expertise in those materials is thin domestically. It's the part of the reshoring story that gets less attention than the fab announcements, and it's the part that could limit actual resilience. One forward-looking signal to hold. Intel's October hike is the first wave. TSMC is expected to raise its own prices in January 2027 When that lands at cascades, AMD, Nvidia, Qualcomm and Intel consumer products all absorb the input cost and propagate it to retail. That's a potential second wave of price increases in consumer hardware in the first half of 2027 The cumulative picture is two years of structural price pressure on hardware, driven by a single upstream allocation decision. AI gets the memory, consumers pay the difference. The watchpoints between now and then are whether AMD holds its current TSMC cost advantage after the January hike, and whether Micron's US ramp actually delivers on the 2028 relief timeline. Those two data points will determine whether the consumer hardware market stabilizes or extends the squeeze further. Thanks for listening. This podcast was built using AI technology, a Yes We production.