May 20, 2026 The defining challenge in advanced semiconductor innovation today is aligning breakthrough ideas with the realities of manufacturing as early as possible. As architecture becomes more interdisciplinary and system‑driven, early technical decisions increasingly shape what can ultimately be built, produced and sustained over time. When manufacturability, yield and system level integration are addressed late, options quickly narrow and iterating becomes expensive. This dynamic is reshaping how new hardware platforms are conceived and funded. Advances now span materials, packaging, devices and systems, extending the distance between research insight and manufacturable silicon while raising both technical and capital risk. Success in this environment depends less on proving that something works in principle and more on ensuring that foundational choices are made with scale and durability in mind. As a result, the next generation of category‑defining platforms is emerging from founders and investors who engage manufacturing realities early, think across the stack and commit to long-term collaboration. These shifts are expanding the role of startups within the foundry ecosystem, positioning them as central contributors to how the next era of computing is designed and industrialized. The hardware innovation cycle is different this time For decades, innovation in semiconductors was largely paced by transistor scaling. Today, differentiation increasingly occurs above the device level, through system architecture, integration strategy, packaging and the combination of diverse technologies on a single platform. As leading-edge CMOS advancement concentrates among fewer players, competitive advantage is shifting toward the ability to reliably orchestrate complex systems at scale. This evolution is driving a renewed wave of hardware startups. Innovation is increasingly defined not by individual technologies, but by how systems come together to serve distinct end markets. In data center infrastructure, startups are advancing silicon photonics, optical connectivity, GaN power and new compute approaches to scale performance and efficiency. In Physical AI, innovation is accelerating across sensing, compute and memory, including emerging memory and neuromorphic architectures. These are not linear transitions from research to deployment, but areas of intense innovation where integration and manufacturability will determine what ultimately scales.Foundries are no longer just endpoints for innovation. They are platform enablers. And startups that were once viewed as peripheral to manufacturing are now central to how the next generation of systems will be built and scaled. Where momentum is forming—and why durability matters For investors, startups and technology leaders, incremental gains no longer cut it. Platforms that not only survive but thrive are designed to scale from day one. The markets attracting sustained capital and attention are those where correctness, integration and manufacturability determine success, not how fast a first demo appears. These are long‑cycle bets with real technical risk, but they are also the domains where enduring value is created. What separates winners from stalled experiments is not ambition, but timing of hard decisions. In deep‑tech hardware, risk is rarely eliminated—it migrates. Too often, manufacturability, yield and system‑level constraints emerge only after capital has been deployed and architectures have hardened. At that point, learning becomes expensive. Late discovery drives redesigns, compresses schedules and erodes capital efficiency for both founders and investors. The advantage belongs to teams that confront these constraints early. Early alignment does not limit innovation. It expands by preserving optionality when trade‑offs are cheapest and most consequential. This is where the right partners become a structural advantage, transforming durability from a liability into a differentiator. How GF approaches startups GF approaches startups as long‑term partners, and in many cases, as the originators of future platforms that will shape the industry’s direction. Many of the most important technologies of the next decade will emerge outside established roadmaps. Engaging early allows those ideas to mature with manufacturing realities in mind rather than being retrofitted later. Our startup engagement model is intentionally flexible. In some cases, collaboration leads to acquisition when capabilities strengthen our core portfolio or accelerate entry into strategic markets. In others, we pursue licensing, direct investment, or deeper partnership to accelerate innovation, create long‑term demand for differentiated platforms and gain early insight into emerging architectures. Not every engagement looks the same, but each is intentional. This philosophy reflects a belief shared by successful founders and investors alike: in complex hardware markets, platforms matter more than products, and partnerships matter more than transactions. GF Labs and Accelerate: two pillars of startup engagement GF’s engagement with startups follows two complementary tracks, reflecting the different ways innovation enters the semiconductor ecosystem. Some startups build differentiated products using proven GF technology platforms—for example, developing architectures or applications on 22FDX. Others are advancing foundational physical sciences innovations, where the underlying technology itself is still maturing. As our research and development model, GF Labs combines internal R&D with a strong external ecosystem of startups, universities, research technology organizations (RTOs), equipment partners and EDA vendors to engage startups early in the maturity cycle of their technology. This early engagement helps ensure emerging innovations follow a clear path to high-volume manufacturing, “future-proofing” key decisions around materials, devices and integration before they become restraints, even when commercial pathways are not yet fully defined. GF Accelerate extends this model into capital. As our venture investment program, it aligns long-term capital with the engineering and manufacturing expertise required to turn breakthrough ideas into industrial-scale reality. Engaging early helps ensure startups are not only well-funded but building technologies that can scale—bridging the gap between what works in the lab and what can be manufactured at volume. Together, GF Labs and Accelerate give founders a path from earliest-stage research through capital formation and into manufacturable silicon. Building ecosystems deliberately No deep‑tech startup succeeds alone. The most effective innovation ecosystems intentionally connect research, manufacturing, capital and enablement. Many promising startups originate as university or consortium spin‑outs that are rich in technical insight but early in their exposure to manufacturing scale. Through GF Accelerate, we partner with experienced venture leaders rather than going it alone — participating as a limited partner across leading platforms in the U.S., Europe and Singapore. That ecosystem includes Silicon Catalyst and Socratic Partners, Cloudberry — Europe’s first dedicated semiconductor, photonics and advanced materials fund — and, most recently, Playground Global, whose Fund IV invests across compute, automation and life sciences. Each partnership deepens GF’s connection to founders tackling the foundational technologies behind AI data centers and Physical AI. These are not passive investments. They allow us to partner closely with founders and investors, support early‑stage companies and create tighter feedback loops between research, capital formation and manufacturing execution. Turning innovation into credible silicon For startups and their investors, the path to silicon is where risk concentrates and where partnership matters most. GF works directly with startups to help mature technologies and deliver key proof points, leveraging the broader ecosystem where appropriate and supporting the transition into manufacturing at the right stage. We provide access to critical resources including process design kits (PDKs), multi‑project wafer (MPW) programs, foundation IP, reference designs, infrastructure and direct engagement with GF technical experts. This approach helps startups de‑risk execution, accelerate learning and build confidence that breakthrough ideas can scale. Looking ahead The next era of computing will be defined not by incremental advances alone, but by foundational shifts in materials, architectures and system integration. Startups will play a central role in this disruption, particularly where challenges extend beyond the reach of any single organization. GF is committed to building alongside founders, investors and technology leaders who share a long-term view of value creation. If you are developing what comes next and looking for partners who understand both ambition and constraint, we believe the most enduring success stories are built together.
