Chip Stocks Drop on Alphabet Blowback — But SK Hynix Tells a Different Story
Intel fell 4%, AMD dropped 3.8% on Alphabet's earnings miss. But SK Hynix's sevenfold profit surge signals AI memory demand through 2030.
Key Points
- Intel dropped 4%, AMD fell 3.8%, and Broadcom lost 3.2% Thursday morning as Alphabet's $9.07B free cash flow burn spooked investors in hyperscaler-dependent chip suppliers.
- SK Hynix reported sevenfold profit growth Wednesday and warned of a structural memory supply shortage heading into 2030 as AI inference and physical AI applications accelerate demand for high-bandwidth memory.
- Advantest raised its FY2027 operating profit forecast 14.3% to ¥967B on 117% net profit growth, confirming that the AI chip testing supply chain remains in full expansion mode regardless of Thursday's spot selloff.
Intel fell 4%, AMD dropped 3.8%, and Broadcom lost 3.2% in Thursday morning trading — not because anything changed in the semiconductor demand picture, but because Alphabet burned $9.07B in free cash flow and the market punished every supplier in its orbit. Twenty-four hours earlier, SK Hynix reported a sevenfold profit surge and forecast a structural memory shortage through 2030. Both facts are true simultaneously, and the tension between them is the chip trade right now.
Why Alphabet's Miss Hit the Wrong Stocks
The semiconductor sector's Thursday morning selloff is a sentiment contagion, not a fundamental revision. Alphabet's negative free cash flow and EPS miss rattled investors who immediately questioned whether hyperscaler AI capex commitments are sustainable — and chip stocks, which have been pricing in sustained infrastructure spending, got caught in the crossfire. But the causality here is worth examining carefully, because Intel, AMD, and Broadcom did not report disappointing earnings Thursday. They reported nothing. They fell because one of their largest customers had a bad quarter driven primarily by mark-to-market losses on a private equity portfolio, not by any reduction in AI infrastructure spending.
Alphabet explicitly maintained its full-year capex forecast of $195B–$205B during the Q3 call — it is not pulling back. The company raised $28.7B in senior notes and $12.4B in equity during Q3 specifically to finance that build-out. The capex is happening. The data centers are being built. The GPUs, custom accelerators, and memory are being ordered. What changed Thursday morning is investor tolerance for holding chip suppliers through a period of hyperscaler earnings volatility — not the underlying demand signal. That distinction matters enormously for traders deciding whether to buy the Thursday dip or interpret it as the beginning of a structural re-rating.
Broadcom's 3.2% decline deserves particular scrutiny. The company derives significant revenue from custom AI accelerator programs — specifically its TPU work with Google — and from networking silicon that goes into hyperscaler data centers. If Alphabet is continuing to spend $195B–$205B in capex, Broadcom's order book does not shrink because of one bad FCF quarter. The stock's Thursday move is a technical reaction, not a fundamental one, and traders with a 60-to-90-day horizon should be treating it accordingly.
SK Hynix's Forecast Is the Structural Signal That Survives Thursday
While the AI chip sector was absorbing Alphabet-related selling pressure Thursday morning, the memory market was telling a completely different story — one that has not changed and will not change because of a single hyperscaler's quarterly income statement. SK Hynix reported Wednesday that profits increased sevenfold last quarter, driven by explosive demand for high-bandwidth memory used in AI training and inference chips. More importantly, the South Korean memory giant issued a forward-looking supply warning: it sees a structural shortage in memory heading into 2030 as AI inference workloads scale and adoption migrates from cloud data centers into physical AI applications — robotics, autonomous driving, and edge AI devices.
That warning triggered a sector-wide rally Wednesday that put the Thursday selloff in proper context. Micron jumped 4% Wednesday, Samsung surged 9%, Applied Materials gained more than 2%, and Lam Research was up more than 2% — all on the strength of a memory demand signal that extends four years forward. One day later, those same stocks are giving back a portion of those gains on Alphabet's unrelated income statement problem. The HBM shortage thesis is intact. The capex cycle is intact. The Thursday selloff is noise laid on top of a genuine multi-year structural signal.
Advantest's Thursday earnings confirm this reading with hard numbers. The Japanese chip testing equipment maker raised its fiscal year 2027 operating profit forecast by 14.3% to ¥967B and lifted sales guidance to ¥1.863T, citing surging demand for testing equipment used on AI-focused high-performance chips. First-half net profit rose 117% to ¥368.1B on revenue growth of 60.5%. Shares surged approximately 9% Thursday. Advantest's raised guidance is the most underappreciated data point in the chip sector this week — because testing equipment demand is a leading indicator of chip production volumes, and those volumes are clearly not slowing down.
Intel's Restructuring Story Needs a Customer, Not a Roadmap
Intel's 4% Thursday decline is the most complicated move in the semiconductor sector, because unlike AMD or Broadcom, INTC has genuine company-specific headwinds that are independent of Alphabet's cash flow problems. CEO Lip-Bu Tan told investors Thursday that Intel has completed approximately 80% of its restructuring efforts — expense cuts, organizational overhaul, and manufacturing strategy realignment. The stock fell 4% anyway. The market's message is clear: completion of restructuring is not the same as proof of customer commitment to Intel's foundry.
The foundry business is where Intel's long-term value proposition either works or fails. Tan's restructuring has stabilized the cost base and reorganized the engineering teams, but the critical unknown — whether leading-edge logic customers will commit to Intel Foundry over TSMC for future process nodes — remains unanswered. Until a tier-one customer publicly commits production volume to Intel's 18A or subsequent process nodes, every restructuring update is a roadmap story rather than a revenue story, and the market has been very clear that it is no longer paying for roadmaps in the AI era.
The Sandisk development Thursday offered a brighter contrast within the storage sector. Sandisk gained more than 4% in premarket after Seagate confirmed that AI and data center demand remains strong, with customer storage commitments locked through 2028–2030. That demand lock-in — customers committing storage purchases three to four years forward — reinforces SK Hynix's shortage thesis and suggests the memory and storage supply chain is entering a period of structural undersupply relative to AI workload growth. For traders, the level to watch in the semiconductor sector is straightforward: if Amazon's AWS print tonight clears the $26B operating income threshold and confirms continued infrastructure expansion, Thursday's INTC, AMD, and AVGO selloff reverses fully within 24 hours. If AWS disappoints, the Alphabet contagion has legs into next week, and the 80-day moving average in the PHLX Semiconductor Index — currently sitting approximately 4% below Wednesday's close — becomes the next technical test for the entire sector.
The Weekly Investor
Daily market analysis for active traders. Free.
