As of 5:45 a.m. ET on Thursday, Dec. 18, 2025, semiconductor stocks are setting up for another volatile session after a sharp, AI-led pullback on Wednesday — but with a major counterweight: Micron Technology’s blowout results and guidance are reviving the “AI hardware” trade in premarket action. The tug-of-war is clear: strong chip demand signals (especially AI memory) versus growing investor scrutiny of AI infrastructure spending, financing, and competition across the stack. Reuters+2Reuters+2
Below is what’s driving chip stocks right now — and what to watch at the open.
Micron’s earnings jolt puts AI memory back in the driver’s seat
The biggest semiconductor headline heading into today’s U.S. session is Micron’s upbeat profit outlook, powered by tight supply and pricing strength in high-bandwidth memory (HBM) — a key component in training and deploying generative AI models. Reuters reported Micron shares were up about 9% in premarket trading after the company delivered a significantly stronger-than-expected quarterly profit forecast.
Why Micron matters for the whole sector today: memory isn’t just “another chip category” in the AI cycle. HBM is increasingly treated as a gating item for advanced AI systems — meaning a bullish Micron print can spill over into expectations for the broader compute supply chain, from GPUs to advanced packaging and wafer-fab equipment.
Key datapoints and commentary investors are reacting to:
- Guidance strength: Reuters highlighted Micron forecasting second-quarter adjusted profit at nearly double Wall Street expectations and pointing to a supply crunch in memory chips amid booming data-center demand.
- Supply tightness beyond 2026: Micron CEO Sanjay Mehrotra said he expects memory markets to remain tight past 2026 and indicated the company may only meet half to two-thirds of demand from several key customers in the medium term.
- Capex escalation: Micron has increased its 2026 capital expenditure plans to $20 billion, reinforcing the view that the AI buildout is pulling forward investment across memory capacity and process upgrades.
- HBM market math is getting bigger fast: Barron’s reported Micron management projecting HBM market growth from $35 billion in 2025 to $100 billion by 2028, alongside warnings that supply-demand imbalance could persist.
Micron’s results also provide a timely contrast to Wednesday’s fear trade: while some investors have started questioning whether AI infrastructure spending is becoming too debt-heavy or too speculative, Micron is effectively arguing that real demand — and real pricing power — remain strong in at least one critical bottleneck category.
The overhang: AI infrastructure financing fears hit chips hard on Wednesday
Micron’s upside surprise is arriving right after a broad, painful move lower in chip stocks tied to AI funding jitters.
On Wednesday, Reuters reported that Wall Street fell to multi-week lows, with chip leaders Nvidia down 3.8%, Broadcom down 4.5%, and the broader Philadelphia Semiconductor Index (.SOX) down 3.9%. The trigger was renewed concern about the cost and funding of AI infrastructure — particularly after Oracle fell 5.4% on a report tied to uncertainty around financing for a $10 billion data center project.
The most important part for semiconductor investors isn’t Oracle’s stock itself — it’s what the episode symbolizes: the market is starting to stress-test the ROI timeline for massive AI spending and the balance sheets supporting it.
Reuters quoted a Baird strategist describing “percolating anxiety” about the AI trade and framing the big question into the new year as the sustainability and return on investment of the spending wave. Reuters
Even this morning, Reuters said U.S. stock futures were firmer (suggesting a possible respite), but the same story underscored that AI spending concerns have resurfaced and remain front-of-mind.
Nvidia’s moat question gets louder: Google and Meta move to reduce CUDA switching costs
Another major semiconductor-adjacent headline with direct implications for Nvidia (and AMD) is a fresh competitive narrative around the AI software stack.
Reuters reported that Google is pursuing an internal initiative, “TorchTPU,” aimed at making Google’s Tensor Processing Units (TPUs) more compatible with PyTorch, the widely used AI framework backed heavily by Meta. The stated goal: reduce friction for customers who want alternatives to Nvidia’s GPUs and the CUDA ecosystem. Reuters
Why this matters for chip stocks:
- Semiconductor winners have been chosen not just by hardware performance, but by ecosystem lock-in. Wall Street analysts frequently cite CUDA as Nvidia’s durable advantage. Reuters explicitly ties the TorchTPU effort to weakening that dominance by lowering software barriers and switching costs.
- Competition isn’t only “another faster chip.” It’s also cloud providers attempting to make alternative silicon easier to adopt at scale — the kind of shift that can influence long-run pricing power and market share assumptions.
This doesn’t mean Nvidia demand disappears — far from it — but it adds another variable for semiconductor multiples: when investors are already nervous about AI capex and debt, stories about rising competition can amplify short-term volatility.
TSMC and Taiwan: export boom supports the AI chip supply chain — tariffs remain a watch item
Overnight from Asia, Reuters reported Taiwan’s central bank raised its 2025 economic growth forecast sharply, attributing strength to booming tech exports to the United States. The report explicitly links Taiwan’s momentum to its role producing advanced semiconductors that power the AI boom for companies like Nvidia, with TSMC central to that ecosystem.
Key details markets are likely to file under “supportive demand backdrop” for leading-edge supply chains:
- Taiwan’s central bank raised its 2025 growth estimate to 7.31% (from 4.55%) and projected growth of 3.67% next year.
- It also noted Taiwan goods face a 20% U.S. tariff, but semiconductors have thus far been excluded, while officials remain in talks with Washington.
For semiconductor stocks, this is a two-sided signal: the AI export boom remains powerful, but trade policy and geopolitics are still a headline risk that can change sentiment quickly — especially for globally exposed names across foundry, equipment, and chip design.
ASML and equipment stocks: China’s EUV “Manhattan Project” adds a new geopolitical layer
On the geopolitics and supply-chain side, Reuters published an exclusive detailing China’s efforts to develop a prototype extreme ultraviolet (EUV) lithography system — a capability currently associated with ASML’s technology leadership and Western export controls.
Reuters reported that a Shenzhen team completed a working EUV prototype in early 2025 and that it is undergoing testing, but has not produced working chips. The report also noted a Chinese government target of 2028 for working chips, while sources said 2030 is more realistic.
This matters for semiconductors stocks in two ways:
- Near term: export restrictions and geopolitical frictions continue to shape the competitive landscape and revenue visibility for global equipment suppliers.
- Long term: even incremental progress on advanced tooling inside China could affect expectations for the duration and enforceability of technology “moats,” especially in lithography and adjacent process steps.
Investors don’t need China to “solve EUV” tomorrow for the story to move stocks — they just need the probability distribution to shift.
The longer-term forecast still points up: SEMI sees record equipment spending through 2027
While today’s tape is being driven by earnings and AI-capex sentiment, the industry’s medium-term spend outlook remains constructive — especially for semiconductor equipment makers.
SEMI’s latest forecast projects global semiconductor manufacturing equipment sales rising to a record $133 billion in 2025, then $145 billion in 2026, and $156 billion in 2027. SEMI attributes the growth primarily to AI-driven investment in leading-edge logic, memory (including HBM), and advanced packaging.
For equipment-exposed stocks — such as etch, deposition, inspection/metrology, and test — this kind of forecast can act as a stabilizer on weak days, even when AI narratives wobble.
What to watch today in semiconductor stocks
Here are the catalysts most likely to shape semiconductor trading as the U.S. session approaches:
1) Micron’s read-through to the rest of AI hardware
- Watch how the market treats Micron’s guidance as a signal for HBM availability, pricing, and allocation — and whether that spills into strength for adjacent chip and equipment names.
2) “AI capex credibility” headlines
- Wednesday’s selloff showed how sensitive chip stocks have become to signs that AI infrastructure spending is getting harder to finance — or harder to justify.
3) Nvidia ecosystem pressure — competition plus software
- Google’s TorchTPU effort is a reminder that the competition story is no longer just about silicon performance; it’s about lowering switching costs for alternative accelerators.
4) Macro rates and risk appetite
- With growth stocks sensitive to yields, traders will be watching central-bank decisions and U.S. inflation data. Reuters also noted Fed Governor Christopher Waller said the central bank still had room to cut rates amid a softening jobs market — a potentially supportive macro counterweight if bond yields move down.
5) Geopolitics and trade
- Taiwan’s export boom is a tailwind, but tariff negotiations and technology restrictions remain live variables. China’s EUV push adds a longer-term strategic wrinkle, particularly for toolmakers and the advanced-node supply chain.
Bottom line for Dec. 18: a classic “two-narrative” semiconductor session
Semiconductor stocks are entering today with two powerful narratives pulling in opposite directions:
- Bull case (demand and pricing): Micron’s results and guidance reinforce that parts of the AI chip supply chain — especially HBM memory — remain tight, high-margin, and demand-constrained.
- Bear case (sentiment and sustainability): investors are increasingly focused on whether AI infrastructure spending is becoming too debt-heavy, too circular, or too competitive to sustain current valuations — a view that hit Nvidia, Broadcom, and the broader chip index hard on Wednesday.
With premarket indicating a bounce in at least some chip names after Micron’s print, the key question today is whether the market treats Micron as “proof of durable AI demand” — or as a temporary exception in a sector still grappling with capex skepticism and competition.
