With so much investment focus on R&D, chip innovation is accelerating. “Look at Taiwan Semiconductor Manufacturing Company (TSMC): moving from 28nm to 10nm took seven years. But now we are seeing a new process node – from 5nm to 3nm to 2nm – announced every one or two,” says William Cheng, senior vice president at iST.  

2025 is the 60-year anniversary of Moore’s law, and transistors are now nearing atomic dimensions. The next leap is new architectures. “Since we have hit physical limits on shrinking, innovation is moving to the back end. Smart packaging – heterogeneous integration – is the next frontier,” says Burkhardt Frick, CEO, SUSS MicroTec. Heterogeneous integration is moving fast too, propelled by the economic value of AI systems; that’s why Samer Kabbani, CEO of AEM Holdings, calls the commercial sweet spot “the middle stack – package assembly and integration testing.” 

Meanwhile, new players are joining the game: In February, Arm announced it would start building its own chips - a move that will bring it into direct competition with many of its customers. Raspberry Pi has also begun selling chips standalone and for the first time, expects to sell more chips than boards this year: “The interesting journey is becoming a semiconductor company,” says Dr. Eben Upton CBE, Founder & CEO. “Over the next decade, we will probably become a company with co-equal electronics and semiconductor businesses.”

The nascent field of photonics offers an especially exciting approach to new chip and server architectures. Computing with light (photons) is faster, carries more data and dissipates less heat than using electrons. In July, Stuttgart-based photonics company Q.ANT GmbH switched on the world’s first operational photonic processor. “Across the U.S., Europe and China, patent activity in photonics is neck and neck. Everyone knows the computer ecosystem must change in the next five to seven years,” says CEO Michael Förtsch. Will these low-temperature systems threaten ROI for data center operators who have invested in expensive liquid cooling technologies? Prof. Dr. Mohr does not think so: “Photonics is certainly disruptive, but it will require significant further development and investment before it can be industrialized and commercialized. Over the next three years, I do not see it as a major disruption,” he says. And for all its promise, photonics is not a panacea. “For now, photonics should be viewed as a complementary technology rather than a comprehensive solution,” cautions AEM’s Kabbani.