The economic and manufacturing challenges of creating monolithic chips is driving an evolution toward chiplet-based designs: small, functional blocks of silicon, each optimized for a specific task and combined within a single package to create a complete system-on-chip (SoC). Because less critical components can be built on mature, cost-efficient nodes, while cutting-edge processes can go to high-performance blocks, there are scalability and performance gains, as well as superior manufacturing yields, with less waste and cost. For these reasons, AMD, Intel, NVIDIA, AWS, Arm, and others are moving toward chiplet-based architectures, with expectations for other innovations, like multi-die packaging on the horizon.  

Those efficiency boosts and cost savings are what hyperscalers, data center providers and engineering firms increasingly need with massive data centers and large-scale “AI gigafactories” in the works, like those now planned across Europe by The European Commission, the European Investment Bank, and the European Investment Fund. In a memorandum of understanding, all three entities agreed to strengthening collaboration and financing for projects aimed at making Europe “a leading AI continent.”  Plans include three to five high-performance computing hubs across the continent, each of which will host roughly 100,000 AI chips to train the latest and most complex models. The facilities will be funded through InvestAI, a €200 billion ($209 billion USD) investment framework that includes a €20 billion fund ($23.32 billion USD) specifically dedicated to the gigafactory initiative.

As massive data centers and gigafactories are constructed for the sake of AI, Editor Sean Kinney of RCR Wireless asked Rob Knoth, an engineer and senior group director of strategy and new ventures for Cadence, is AI helping or hurting those who engineer and build the actual facilities? Knoth said: “When I talk about that affliction of being an engineer or a designer, what it really comes down to is that intuition, that imagination. It’s that grounding in a very solid understanding of [computer science] and math, of really understanding those fundamentals. When you couple that with intuition and with imagination, that’s the spark. That’s the beauty.” He acknowledged engineers’ roles continue to shift toward integrated, strategic, and tech-driven oversight, with mastery in skills like automation, sustainable design, and digital modeling. He emphasized engineers will continue to embrace change and automation, even when things aren’t quite as “pretty, or quite as good” as they’d be with a human doing them. There’s an acceptance that “if the computer, the software, the tool can do something that meets the requirements that you’re measuring, and it can do it in a semi-automated fashion, you learn to trust it.”