Home/Research/European Semiconductors: The Cycle Bottom Is Real, but the Recovery Is Priced
中文
May 26, 2026 阅读中文版

European Semiconductors: The Cycle Bottom Is Real, but the Recovery Is Priced

Executive Summary

European semiconductors are being repriced in a subtle way. The cycle bottom is real, but the equity market has already moved far ahead of the financial recovery. STMicroelectronics is the cleanest case study. In 2025, the company had just completed a deep downturn: revenue fell to $11.75 billion, net income dropped to $166 million, and net margin compressed to 1.5%. By the first quarter of 2026, revenue was up 23% year over year, management was again talking about AI data-center power, and the stock had moved from its March low to near a 52-week high by late May.

This is not a single-stock note. It is a broader question about European semiconductors: when an old growth engine breaks and management replaces it with a new narrative, how much credit should the market give? Across STM, Infineon, NXP, and onsemi, the answer is uneven.

The Cycle Bottom Is Visible

Start with the positive facts. STM reported first-quarter 2026 net revenue of $3.10 billion, up 23.0% year over year, gross margin of 33.8%, and net income of $37 million. For the second quarter, management guided to roughly $3.45 billion of revenue and gross margin of about 34.8%. Idle capacity charges were also declining, which suggests inventory digestion and utilization recovery have started.

These are the signals that usually trigger the first leg of a cyclical recovery: revenue stops falling, gross margin stops deteriorating, and management begins to talk about better bookings. The problem is that equities often move faster than reported earnings. STM rose from roughly $30.29 on March 20, 2026 to about $66.86 on May 22, a gain of roughly 121% in two months. At that point it no longer looked like an ignored cycle bottom. It looked like a recovery story with the narrative already rebuilt.

The broader European power-semiconductor backdrop is similar. Demand has not vanished. Automotive electronics, industrial control, and data-center power remain real markets. But the old electric-vehicle SiC supercycle has been repriced. The old story was that EV penetration would rise steadily, main inverters would use more silicon carbide, and European integrated device manufacturers would enjoy high-margin growth. The current reality is slower EV growth, faster Chinese self-supply, and a clear signal from Tesla as early as 2023 that next-generation platforms would use materially less SiC.

The cycle bottom is real. It is not a return to the old 2021-2023 curve.

The Structural Ceiling Is Lower

The key issue for STM is gross margin. Gross margin was 47.94% in 2023, 39.34% in 2024, and 33.89% in 2025. A 33.9% gross margin is weak among large peers: NXP is around 54.7%, Infineon is around 41.4% on an adjusted basis, and onsemi is around 33.1%. STM has shifted from being viewed as a high-quality European IDM to looking more like onsemi in the stressed power-cycle bucket.

That is not just normal cyclicality. The pressure has two layers.

The first is product mix. NXP’s automotive processors, MCUs, and radar assets have been more resilient. Infineon’s power and security-MCU mix has also held up better. STM and onsemi have heavier exposure to SiC and power devices, so when the EV and SiC story cooled, their margins fell more sharply.

The second is competition. Low-end MCUs and some power devices are facing price pressure from Chinese suppliers, and European companies cannot assume that historical ASPs will return simply because demand recovers. Even if volumes improve, margin may not return to the old highs.

That is the difference between a cyclical recovery and a structural repair. A cyclical recovery means revenue and utilization are moving upward. A structural repair requires the company to prove margin, product quality, customer stickiness, and pricing power. STM has proved the first. It has not yet proved the second.

AI Power: New Engine or New Packaging?

Management is now emphasizing AI data-center power. In its first-quarter 2026 results, STM confirmed that it expects data-center revenue to be well above $500 million in 2026 and well above $1 billion in 2027. This is not an empty direction. AI servers need higher power density, higher efficiency, and more advanced power-management components, including silicon carbide and gallium nitride. European power-semiconductor companies have real capabilities here.

But three questions need to be separated.

First, $1 billion is meaningful, but relative to a potential $14-15 billion revenue base, it is still a mid- to high-single-digit contribution. It is not an immediate company-level transformation.

Second, this is a projection, not a public take-or-pay contract or a customer-level backlog disclosure. STM has changed long-term targets before. Its 2024 capital markets day pushed the $20 billion revenue ambition from 2027 to 2030. The market can assign some value to a new narrative, but it should not treat a projection as already contracted cash flow.

Third, every peer is telling an AI-power story. Infineon, NXP, and onsemi are also emphasizing data-center power, physical AI, and power efficiency. When an entire industry pivots to the same narrative, the opportunity is real, but it is not unique to one company.

AI power is therefore a candidate for the next product-mix repair cycle. It is not yet a proven second growth curve.

Peer Comparison: Recovery Versus Narrative Swap

The peer set makes the split clear.

NXP has a gross margin around 54.7%, a net margin around 16.9%, and a two-year cumulative revenue decline of roughly 7.6%. Its business is more weighted toward automotive processors, MCUs, and radar. It has cycle pressure, but the profit structure is more intact. NXP can be discussed on current earnings.

Infineon’s adjusted gross margin is around 41.4%, and its profit structure has also held up better than STM’s. It has power and SiC exposure, but its revenue decline has been shallower, which points to a more resilient product and customer mix.

STM and onsemi are the stressed pair. STM’s two-year cumulative revenue decline is about 32%, and onsemi’s is about 27%. Both have gross margins around 33%, and both have seen net margins compressed to low levels. Their trailing earnings multiples are not very useful because the denominator has collapsed. Investors have to believe future earnings will recover before accepting today’s price.

Using NXP as the anchor, STM is not obviously cheap. NXP has strong current margins now. STM needs a 2027 recovery before it approaches a forward multiple around the high 20s. In other words, the market has already capitalized the recovery.

Europe’s Structure: State Ownership Reduces M&A Optionality

European semiconductors also carry a structural feature that US investors often underweight: strategic state ownership.

STM has French and Italian state-linked ownership through FT1CI at roughly 27.5%. Soitec also has French state-linked shareholders and enhanced voting-right structures. This does not make the companies bad. It does reduce one common downside assumption: that if valuation falls far enough, someone will acquire the asset.

In the US, cyclical drawdowns often come with activist, private-equity, and strategic-acquirer optionality. In European semiconductors, that optionality is less natural. Strategic-asset status makes takeover premia, breakups, and governance arbitrage harder to underwrite. Valuation work should not simply import a US-style “someone will buy it” floor.

What to Track

First, STM’s second-quarter 2026 results. The important question is not only whether revenue continues to recover, but whether gross margin clearly moves toward the high 30s and whether data-center revenue becomes measurable rather than just projected.

Second, AI-power design wins. If data-center power remains mostly narrative, valuation support is weaker. If disclosures become clearer on customer, platform, revenue size, and lifetime, the narrative quality improves.

Third, peer-margin separation. If NXP and Infineon keep structurally higher margins while STM and onsemi remain in the low 30s, the market will split “cyclical recovery” into “product-quality hierarchy.”

Fourth, the euro. STM has a large European cost base and dollar-denominated revenue. A stronger euro can pressure gross margin. FX is not the thesis, but it matters more when margins are already low.

Conclusion

European semiconductors are not without recovery. The cycle is bottoming, and AI data-center power gives power semiconductors a new demand vector. But recovery and undervaluation are not the same thing.

STM’s problem is that the old SiC engine has been damaged, the new AI-power engine has not yet proved itself with backlog disclosure, and the stock has already moved from cycle-bottom pricing to narrative-complete pricing. More broadly, the next phase of European auto and power semiconductors will be determined by product mix and customer qualification, not simply by a rising semiconductor cycle.

The recovery can be real without deserving the old margin ceiling.

Data and Sources

This report is an independent KSINQ market observation for informational purposes only. It is not investment advice. Data snapshot: May 26, 2026.