While France’s AI Summit in early February brought a wave of massive investment announcements — particularly in new data center infrastructure — the country’s entire microelectronics industry may stand to benefit from this renewed momentum.
The AI Action Summit in Paris became a platform for headline-grabbing commitments, across various domains: nearly €200 billion in public and private investments were pledged at the European level, showing that the 27 EU countries are determined to stay competitive with American and Chinese tech giants who have already embraced generative AI. In France, this ambition translated into President Emmanuel Macron’s announcement of up to €109 billion in public and private investments, including €99 billion earmarked for the creation of 35 new data centers.
“We know that upcoming data centers will rely heavily on well-known platforms — Nvidia GPUs, Intel and AMD CPUs, and integrators like Dell or HP. At first glance, this leaves little room for French players,” notes Damien Bretegnier, Investment Director at Supernova Invest.
But calling it a done deal would be premature for local champions such as Soitec (a silicon wafer supplier based in Isère), STMicroelectronics, and the broader deeptech ecosystem (especially in the Auvergne-Rhône-Alpes region), which includes rising stars like Scintil Photonics, Kalray, and Wise Integration — all of whom stand to benefit from the current AI wave. While precise estimates vary, the opportunities are starting to take shape.
According to Pierre Cambou, a semiconductor industry analyst at Yole Développement, up to half of the pledged investment could end up flowing into microelectronics.
Data Center Growth Driving New Demand
Next-generation data centers will dramatically increase demand for high-power electronics that can manage enormous energy loads with minimal losses. This is where regional giants like STMicroelectronics and Soitec are already positioned, with innovations such as silicon carbide (SiC) — a material that withstands high temperatures and boosts efficiency.
“One of the ways to lower energy consumption in data centers is by using smart materials like silicon carbide, which can improve efficiency by 10 to 20%, and eventually gallium nitride (GaN) near the GPU nodes,” explains Christophe Maleville, Chief Innovation Officer at Soitec.
“We’re already working at our plant in Catania, Italy, on new generations of products like silicon carbide and gallium nitride, aiming to capture 10% market share by 2030,” adds Philippe Magarshack, VP of Strategy for Microcontrollers, Digital & RF at STMicroelectronics.
Data centers will also drive up demand for optical communication components, essential for the fast transfer of data between GPUs. “It’s not enough to have fast GPUs — communication must be equally fast. That means bringing fiber optics closer to compute units,” says Damien Bretegnier, citing Scintil Photonics and CEA-Leti as local players developing technologies that could hit the market within 2–3 years.
“To deliver high performance inside data centers, you need massive bandwidth and data flow. That’s why the industry is moving toward silicon photonics,” adds Maleville. Soitec’s silicon wafers are already key to many AI-focused components.
STMicroelectronics recently announced a partnership with Amazon Web Services (AWS) to supply silicon photonics technologies that will boost AWS’s data center bandwidth.
“This technology will be produced at our Crolles facility (Isère) on 300mm silicon wafers, starting at the end of the year,” confirms Magarshack. The company estimates the market to reach $2 billion by 2030, with 20% annual growth.
The Promise of Edge AI
But data centers are not the only opportunity AI brings. While the Paris summit spotlighted generative AI and advancements from players like ChatGPT and Deepseek, the industry is also turning to “edge AI” — intelligence embedded locally into devices like cameras, vehicles, and robots.
The dual promise? Lower power consumption for more sustainable products, and enhanced security by reducing vulnerability to online attacks.
“Mistral AI is already working on large models with 125 billion parameters, broken into sub-models of 38 billion parameters — a size that’s nearing the threshold of embedded compatibility. That’s good news for both the planet and data security,” adds Damien Bretegnier. He emphasizes that the French ecosystem — along with European neighbors like Infineon and NXP — is well-positioned to lead in low-power technologies that will drive embedded AI growth.
Toward a New Growth Engine for the Industry
Could this new wave of AI reenergize an industry recently hit by slowdowns in smartphoneand EV sales?
“What’s interesting is that the AI surge comes at a critical moment for European industry — after post-Covid supply shortages, and now a pricing downturn. We’re seeing a return to normal in the supply chain,” says Cambou. “But that doesn’t mean demand has vanished: EV needs remain, and now data centers are joining the mix.”
“We’re currently in a downturn, but it’s precisely in these times that well-run companies invest in the future — especially in next-gen AI technologies. Semiconductors will play a central role in this momentum, even if the impact takes a few years to materialize,” adds Magarshack.
Soitec also expects AI — via its silicon photonics and edge computing divisions — to spill over into other markets.
“AI now touches every sector: automotive, smartphones, and software platforms like ChatGPT and Deepseek. It’s becoming harder to draw boundaries,” notes Maleville. He believes AI innovation will rekindle consumer demand through technological breakthroughs.
In a sector that typically reinvests 20% of annual revenues in R&D, recent European efforts — such as the Important Projects of Common European Interest (IPCEI) with €8 billion in public funding — aim to support the development of next-gen chips by 2032.
“In 2024, semiconductor growth was driven only by cloud. The rest of the segments declined. But firms like STMicro appear to have bottomed out. According to analysts like Jefferies, growth should resume next year — powered by both cloud AI and edge AI,” concludes Damien Bretegnier.
To read the original piece: https://region-aura.latribune.fr/innovation/2025-02-25/pourquoi-les-milliards-de-l-ia-sont-enfin-une-bonne-nouvelle-pour-la-microelectronique-francaise-1018831.html
