Applied Materials to invest $5 billion in AI chip R&D center

Applied Materials has unveiled the EPIC Center, the largest semiconductor R&D center in US history. The $5 billion investment reflects the scale of the challenge: the energy efficiency of AI systems now depends not only on raw computing power, but also on how efficiently data moves between logic and memory.

Why the old R&D model no longer works

For decades, the semiconductor industry developed chips like a relay race: one team created components and passed them on through integration and manufacturing, and only then did engineers from other disciplines give feedback. That model worked as long as progress depended on separate, independent steps.

In the AI era, that has changed. At angstrom scale (billionths of a meter), physics forces every part of the system to be tightly linked. Material choices affect integration, integration sets design rules, those rules determine power delivery, and wiring sets the thermal budget. System architects cannot wait 10–15 years between cycles of technological innovation.

Three areas of development

The EPIC Center focuses on three interconnected areas:

• Logic: a new type of transistor (GAA) with improved energy efficiency
• Memory: 3D DRAM with higher capacity and lower power consumption
• Packaging: hybrid bonding and HBM (High-Bandwidth Memory) to bring memory closer to the processor

In logic, Applied Materials is developing gate-all-around (GAA) transistors, three-dimensional devices that increase density without excess energy loss. GPUs now in development already contain more than 300 billion transistors connected by 2000 km of wiring.

In memory, the focus is on the shift from planar structures to 3D DRAM, which stacks memory cells vertically to increase capacity within the same die area.

In packaging, the key technology is hybrid bonding, which connects components at a density close to that of wiring inside a chip, along with high-bandwidth memory (HBM) for fast access to large volumes of data.

How EPIC changes the process

Unlike the traditional model, EPIC works on a parallel co-innovation model. Customers' engineers work alongside Applied Materials technologists from day one. Modeling, testing, process development and quality control are integrated in a single secure environment. Problems that used to surface in the final stages are now found and solved early.

The result is a path twice as fast, which benefits the whole ecosystem under one roof. Chipmakers get early access to Applied's R&D portfolio, faster learning cycles and technology transfer into production. Ecosystem partners get early access to advanced manufacturing technologies.

What this means

The semiconductor industry is moving from sequential innovation to parallel innovation. The EPIC Center is an attempt to shorten the time from idea to volume production at a moment when AI demands ever more energy-efficient and powerful chips.

For consumers, this means the next generation of AI accelerators and neural network processors will be developed and brought to market faster.