1 week Ago
GlobalFoundreis GTS 2024 AI Everywhere talk by Corporate Fellow Ted Letavic. As we (TIRIAS Research) have written about previously, semiconductor supplier GlobalFoundries (GF) announced in early 2018 that the company was dropping its plans to develop a leading-edge 7nm process technology and beyond. But that does not mean GF has stopped innovating.
At the recent GlobalFoundries Technology Summit, the company continues to invest in advancing mainstream semiconductor process nodes with new materials and advanced wafer processing. In doing so, it has developed unique process platforms that serve the diverse chipmaking needs for a wide variety of customers and for specific market segments. One particular segment is government agencies.
GlobalFoundries is currently the only commercial high-volume foundry with the “Trusted Foundry,” designation. As such, its technological roadmap has an impact on government chip development programs and helped secure CHIPS Act funds. At the GlobalFoundries Technology Summit 2024, the company brought together its customers and partners from around the world.
The theme was “AI Everywhere” as AI has impacted everything from IoT to the datacenter. However, the focus was on its core markets: aerospace, defense and critical infrastructure, automotive, communications and data center infrastructure, industrial IoT, and mobile devices. RF One key technology for many of those markets is RF (Radio Frequency) chips for mobile devices, communications infrastructure, and aerospace.
GF’s mainstream RF offerings use Silicon on Insulator (SOI) wafer technology. But GF also has Silicon Germanium (SiGe) and Gallium Nitride (GaN) processes as well. With this level of diverse RF technologies, the company is well positioned for future 5G NR cellular developments and beyond, including the growth of FR2 (above 24GHz ) and FR3 (7.
1GH to 24GHz) bands. Trump’s Cabinet: Here Are His Picks For Key Roles—Robert F. Kennedy Jr.
, Matt Gaetz, Tulsi Gabbard And More Netflix’s Best New Movie Is A Surefire Oscar Contender Elon Musk Issues Surprise Crypto Endorsement Amid $3 Trillion Bitcoin And Dogecoin Price Boom Most of the GF RF capacity runs on 300mm wafers, and GF will be moving SiGe from 200mm to 300mm in 2027 for higher volume. With RF, GF is pushing performance and will use 3D technologies, like wafer-to-wafer bonding. With wafer-to-wafer hybrid bonding, GF RF SLATE technology can stack transistor in the vertical (Z) dimension, increasing overall transistor density without changing process node.
The newest RF technology for GF is GaN. With GaN you can deliver higher power in smaller form factors. GF announced a strategic technology development and licensing agreement with Finwave Semiconductor, combining Finwave’s intellectual property and GF’s volume manufacturing capabilities.
Finwave's GaN-on-Silicon enhancement-mode MISHEMT platform delivers excellent gain and efficiency at sub-5V voltages for RF chips. GF will use Finwave’s technology on its 90RFGaN platform. The target markets for it include new higher frequency 5G FR2/FR3 bands, 6G and mmWave amplifiers, and high-power Wi-Fi 7/8 systems, where superior range and efficiency are critical.
The companies expect mass production in GF’s Vermont fab in the first half of 2026. Power Power is critical to system design – from intelligent edge and mobile to automotive and datacenters. With datacenters expected to consume up to 10-20% of world-wide power capacity, power efficiency is essential.
GF has BCD (Bipolar-CMOS-DMOS) and GaN technologies for power conversion circuits. BCD is good for up to 150V and with deep trench isolation to 200V. It’s used for DC/DC converters for datacenters and EV Battery management.
Another technology that has a lot of potential is power GaN. GF acquired Tagore Technology’s for its proprietary GaN IP portfolio that can support higher voltages. Silicon Photonics Silicon Photonics (SiPh) is an area where GF is pushing the technology with new material and specialized process – pushing to 100GHz bandwidth.
The next gen SiPh should offer twice the bandwidth and transmission power of 1 picojoule/bit. The next market for SiPh will be XPU to XPU connectivity (such as GPU to GPU) where bandwidth and power are important to building efficient AI supercomputers. SiPh requires 3D packaging that places the laser on the silicon wafer.
Partnerships In its mainstream CMOS line, GF has been adding application-specific features. It has 44 platforms and supports analog and image sensors controller. One of GF’s partnership shows you don’t need angstrom sized transistors to build a power-efficient processor – a unique architecture can provide significant benefits.
Efficient Computer partnered with GlobalFoundries (GF) to bring to market a new high-performance embedded processor that it claims is up to 166x* more energy-efficient than industry-standard embedded CPUs. Efficient Computer combines the MRAM and Adaptive Body Biasing (ABB) capabilities of GF's 22FDX platform for low power and long battery life. Efficient Computer is already working with select customers for early access and customer sampling by summer 2025.
Efficient Computer is targeting the industrial IoT, wearable and implantable health devices, space systems, and security and defense markets with its unique IP. For the Automotive, IoT, and Smart Mobile Segment, GF announced a strategic relationship with leading chip supplier NXP. This collaboration leverages GF’s 22FDX process to optimize the power, performance and time-to-market of NXP’s solutions.
GF’s 22FDX chips are to be manufactured in Dresden, Germany and Malta, New York, providing NXP geographically diverse supply for their customers. Quality is critical for automotive where the goal is zero defects and long support. In addition, the automotive market expects to continue to use 40nm MCUs to 2030 and beyond, providing GF a stable market for older nodes.
Research GF also has a lab with 1500 people, works with over 70 universities and is engaged with 80+ startups and incubators. Some of the cutting-edge technology it’s working on includes cryogenic and quantum systems with back gate electrode cryogenic control. It’s investigating quantum dots that control a single electron.
For AI at the edge, there’s work on hyperdimensional computing that assigns a bit stream to each vector. Summary GlobalFoundries demonstrates that innovation in semiconductor manufacturing is not limited to process geometries. Through innovative materials and processes, GF is enabling new technologies in areas like RF, power, and optics that are foundational to consumer, commercial, and government segments.
And while Moore’s Law may be slowing, there is more to enabling the AI era. Tirias Research tracks and consults for companies throughout the electronics ecosystem from semiconductors to systems and sensors to the cloud. Members of the Tirias Research team have consulted for Intel, GlobalFoundries and other companies throughout the semiconductor ecosystem.
.
