Then, suddenly, the stock woke up. More than that: it really took off. Since the beginning of the year, the share price has more than quintupled after hitting a 9-year low in late 2025. This kind of trajectory rarely tells just one story. It speaks simultaneously of yesterday's despair, today's enthusiasm, and, perhaps, tomorrow's excess.
The reason for this surge can be summed up in one word: photonics. Or more precisely: Photonics-SOI, those silicon-on-insulator wafers adapted for the manufacturing of photonic integrated circuits. These components are used to transport information no longer just with electrons, but with light. In data centers fueled by artificial intelligence, where processors must communicate ever faster with each other, this issue has become strategic.
Soitec thus finds itself propelled to the heart of a very powerful market narrative: optical connectivity for AI. And in the stock market, when a long-neglected French industrial company is suddenly associated with Nvidia, data centers and AI infrastructure, the awakening can be shocking.
However, the problem is already there: the market story may have moved faster than the industrial reality.
Why Soitec is coming back from so far behind
To understand the current frenzy, we need to remember where Soitec came from.
The company is not a chipmaker. It produces advanced materials, particularly SOI substrates, used by foundries and semiconductor manufacturers. Its expertise lies in improving the performance of electronic components: power consumption, speed, integration and industrial yield.
For years, the market primarily viewed Soitec through the prism of smartphones, with substrates used in radio-frequency components. However, this engine stalled. The smartphone supply chain underwent a violent inventory correction. Volumes remained under pressure. In automotive, another major outlet, Chinese market weakness and customer caution also weighed heavily.
As a result, Soitec was abandoned by investors. A fine industrial story, indeed, but without visible growth, with margins under pressure, inventories to clear, and reduced visibility. Indeed, this is the perfect cocktail for multiple compression.
Then the market changed its stance:
The OFC conference and announcements surrounding AI highlighted the colossal need for bandwidth in data centers. AI architectures rely on thousands of processors that must exchange gigantic volumes of data. Copper is reaching its limits. Optics are becoming essential.
This is where Soitec returns to the game. Jefferies points out that Soitec's Photonics-SOI wafers are used to manufacture photonic integrated circuits for pluggable optical transceivers used in "Scale Out" — the connections between equipment within data centers. The broker estimates this market is growing at around 30%, driven by strong demand for high-speed optical transceivers.
In other words, Soitec is no longer seen merely as a stock exposed to the smartphone cycle. It is becoming a stock exposed to the physical infrastructure of AI. This is a narrative shift. And in the stock market, narrative shifts can sometimes matter as much as changes in earnings.
Soitec's case
Soitec's first asset is its technological position. In Photonics-SOI wafers used in silicon photonics, the company and its licensees hold a very strong position. In such a specialized value chain, qualified suppliers are not replaced overnight.
The second asset is the AI option value. Today, demand comes mainly from pluggable optical transceivers used in Scale Out. But the market is already looking at the next step: CPO, or co-packaged optics, which involves bringing optics as close as possible to processors or switches.
If this technology is deployed on a large scale in Scale Up — the connections within the densest computing architectures — Soitec's addressable market could change dimension. This is the central bullish argument: Soitec's photonics revenue could, in a favorable scenario, be multiplied several times over.
The third asset is market scarcity. European investors have few pure-play ways to bet on optics for AI. Aixtron, STMicroelectronics, Nokia, and Soitec appear as the main European names exposed to the theme, each at a different point in the value chain.
Finally, Soitec still has an intangible asset that the market can quickly rediscover: technological credibility. The company has already proven it can impose its specialized substrates on major markets. If silicon photonics becomes a de facto standard in certain optical connectivity building blocks, Soitec is well-positioned to benefit.
Reasons for caution
The limitation is that everything will not become photonic overnight.
The most obvious weakness is the current size of the business. Based on available data, about 17% of fiscal year 2026 revenue would come from Photonics-SOI wafers. This is significant, but it is not yet the core of the group. Even 35% growth in this pocket would only represent about 6 percentage points of growth for the overall business.
The rest of the portfolio remains complicated. More than 80% of revenue remains exposed to much less glamorous markets: smartphones, automotive, and other semiconductor applications. The smartphone sector is still in inventory correction, while automotive offers little short-term support. Soitec's recovery therefore does not depend solely on AI. It also depends on the end of a negative cycle in its historical businesses.
Another point to keep in mind: there is no shortage of Photonics-SOI wafers. Unlike lasers, where certain capacity tensions can support prices and equipment orders, Soitec and its licensees have abundant capacity. This limits the idea of an immediate surge in revenue or margins.
Finally, there is the question of the CPO timeline. This is likely the heart of the debate. Jefferies estimates that its adoption in Scale Up will be gradual, with little significant impact on Soitec's revenue in the short term. Pluggable transceivers remain dominant in Scale Out, while broad adoption of CPO will likely take several years.
Put another way: the market is currently buying an attractive option, but this option might not translate into massive revenue for three years or more. In the meantime, one must live with the actual financial statements.
And these accounts are not yet those of a hyper-growth company. According to recent market data, Soitec is expected to be barely profitable for the fiscal year ending March 2027, while the stock is already trading at high multiples: approximately 92 times expected earnings for fiscal year 2028, 6.8x EV/Sales, and 23x EV/Ebitda (Source MarketScreener/S&P Capital IQ, 14 analysts). The market is no longer pricing Soitec as a struggling cyclical, but as a credible AI beneficiary.
Deutsche Bank vs. Jefferies: Two readings of the same turn
What makes the case interesting is that the positive and cautious views do not focus on two different realities. They interpret the same phenomenon differently.
The positive reading, illustrated by Deutsche Bank a few days ago, argues that the market still underestimates the depth of the transition. Soitec would shift from being a smartphone loser to an AI winner. Upcoming news from foundry customers like Tower or GlobalFoundries could confirm the acceleration. And if SOI wafers become a standard in integrated photonic circuits, Soitec's position deserves a further re-rating.
The cautious reading, defended by Jefferies, acknowledges the quality of the positioning but refuses to extrapolate too quickly. Yes, Scale Out is progressing. Yes, Soitec is benefiting from optical transceiver demand. Yes, CPO is an opportunity. But no, this is not yet enough to offset the weakness in smartphones and automotive. And no, CPO adoption does not seem close enough to justify any multiple.
Both analysts could be right, and that is precisely what makes the situation explosive.
What next?
The future will essentially depend on three things.
The first is commercial confirmation. The market will want concrete signals on Photonics-SOI demand. Announcements from Tower, GlobalFoundries, or other players in the optical chain must be watched closely.
The second is the pace of normalization in historical businesses. If smartphones remain depressed longer than expected, photonics will have to carry a disproportionate weight. If, on the contrary, the inventory correction ends and automotive stops deteriorating, Soitec could benefit from a double effect: cyclical recovery and AI re-rating.
The third is the reality of CPO. This is clearly the wild card. If the technology takes hold quickly in Scale Up architectures, the market will have been right to anticipate. If the timeline slips, the stock could find itself hanging on a promise that is too far off.
MarketScreener's recent series on French companies exposed to AI:
