作者：阿富萨特·迪里斯（Afusat Dirisu），航空航天与国防事业部空间与抗辐射能力副总监 

当“阿耳忒弥斯二号”任务时隔50多年首次将宇航员送至近地轨道之外时，这标志着人类探索史上的又一里程碑。但正是得益于那些能在堪称最严酷的环境中依然完美运行的电子设备，此类任务才得以实现。 

现代航天飞行中的导航、通信和生命保障系统均依赖于半导体。半导体行业也正迈向新的前沿。随着人类的探索目标从月球任务扩展到深空探索和卫星星座，各家公司开始探索在太空中制造芯片，利用微重力环境和自然真空条件来提升生产质量。 

尽管“太空制造”半导体技术仍处于起步阶段，但当务之急是确保芯片能够经受住太空严酷环境的考验。 

太空的现实：硅材料的极限测试  

从火箭离开发射台的那一刻起，电子设备便要承受极端且往往难以预料的环境条件：温度可能剧烈波动，辐射水平比地球高出数个数量级，而真空环境还会对材料产生独特的应力。与地面系统不同，一旦硬件部署到位，便再无维修或更换的机会。 

传统的商用芯片——例如用于智能手机、智能冰箱或电动汽车的芯片——并不具备承受此类辐射强度的能力。仅辐射本身就可能引发瞬态故障，随着时间推移导致材料劣化，甚至对电路造成永久性损坏。在漫长的任务周期中，这些影响会不断累积，最终危及系统的可靠性及任务的成功。 

正因如此，航天领域对半导体设计和制造提出了根本不同的要求，其核心在于韧性、可预测性和可信度。GlobalFoundries 凭借我们专为“设计级抗辐射”（RHBD）技术量身打造的技术组合，直面这一设计挑战。 GF 还通过优化掺杂植入或基板材料工程，探索并开发基于工艺的抗辐射（RHBP）技术节点，以进一步提升总电离剂量（TID）和单事件效应（SEE）等辐射性能指标。这些努力使得电子设备能够在持续辐射暴露下可靠运行。 正是这一技术基础，确保了航天系统在发射后很长一段时间内都能无故障运行。  

针对辐射进行设计，以确保堆芯的韧性 

RHBD技术对于确保芯片能够承受严酷的太空环境至关重要。RHBD技术并非仅依赖屏蔽措施，而是将抗干扰能力融入芯片架构之中。 

通过冗余、空间布局感知以及稳健的故障缓解策略等设计技术，RHBD 使半导体即使在遭受辐射干扰时仍能继续运行。这些方法使设计人员能够在错误传播之前预先识别并加以缓解，从而确保关键任务系统保持稳定的性能。 从近地轨道和地球同步轨道到深空探测器，不同的太空任务需要量身定制的方法来平衡性能、功耗和耐用性，但 RHBD 始终是确保芯片在任务生命周期内不会发生故障或能够从故障中恢复的关键。  

为什么安全是不可妥协的要求  

在航空航天和国防领域，安全性与性能同样至关重要。用于太空领域的半导体通常为敏感系统提供动力，包括通信基础设施和国家安全资产。确保这些芯片在值得信赖且安全的环境中制造，对于防止篡改、保护知识产权以及维护任务完整性至关重要。 

作为一家经认证的“可信代工厂”及“通用标准”认证供应商，格罗方德在制造的每个阶段都贯彻了严格的流程、监管和合规要求。这确保了芯片能够安全无虞地交付，这对在高风险环境和关键终端应用中运营的客户而言，是一项至关重要的要求。 

在全球范围内推动差异化制造 

要满足航天级电子产品的需求，还需具备跨技术和跨地域的灵活性。GF的全球制造网络能够提供多样化的半导体解决方案组合，这些方案专为航空航天、国防及其他关键基础设施应用进行了优化。 

GF在全球各晶圆厂提供针对特定应用的晶圆技术，涵盖用于高性能处理、射频集成和节能平台的高端工艺节点。这使客户能够根据具体应用选择最合适的技术，同时确保质量和可扩展性不受影响。 

在美国，GF位于纽约州马尔塔的工厂正通过FDX® FD-SOI、45SOI和FinFET等先进平台扩大产能，支持从安全通信到边缘计算的各类应用。该工厂还正在开发FinFET RHBP技术，该技术融合了Vorago HARDSIL®，以进一步提升我们12纳米FinFET节点的TID和SEE性能。 我们位于佛蒙特州伯灵顿的工厂是GF在功率和射频领域领导地位的基石，其氮化镓（GaN）和高压硅基氮化镓（GaN-on-silicon）技术正为下一代雷达、卫星和自动驾驶系统提供动力。GF位于新加坡和德累斯顿的晶圆厂因符合通用标准（Common Criteria）认证的安全、高完整性制造而享誉国际。  

凭借这一全球制造实力，我们确保客户能够从容应对产品的整个生命周期，并为太空及更远领域的关键任务应用提供所需的卓越性能、可靠性和信任保障。 

通过生态系统合作加速创新  

没有任何一家公司能够独自解决航天领域的挑战。正因如此，格罗方德致力于打造一个合作伙伴生态系统，共同推动基于GF平台的抗辐射解决方案。  

GF的12LP FinFET工艺级抗辐射（RHBP）平台正与诺斯罗普·格鲁曼航天系统公司及Vorago Technologies紧密合作开发，旨在为需要抗辐射解决方案的客户扩展GF的产品组合。该平台采用了工艺改进措施，在保持设计基准电气性能的同时，使设计人员能够利用现有IP，并显著提升抗辐射性能。 

近日，BAE系统公司推出了其“Ascent”号航天器，该航天器旨在通过先进的机动能力（包括补给和高推力推进）支持动态太空作业，同时可搭载多个拼车有效载荷。这项创新的核心在于BAE系统公司与GF的合作。 BAE的12S0和RH12™ Storefronts分别基于GF的45SOI和FinFET平台构建，为开发定制的“设计抗辐射”（RHBD）解决方案提供了交钥匙路径。这种方法将先进商用技术的性能与效率，与太空应用所需的韧性相结合。 

其结果是在紧凑且可定制的设计中，实现了高性能处理、安全连接、低功耗与高可靠性的强大结合。这些生态系统合作伙伴关系正在加速从概念到入轨的进程，助力开发更加敏捷，降低项目风险，并确保新一代航天系统能够在最严苛的环境中稳定运行。 

将当今的需求与未来的可能性相连接  

在太空中制造半导体的愿景，最终可能会重塑整个行业，为性能和材料领域开辟新的疆域。但鉴于这一未来尚在酝酿之中，像“阿耳忒弥斯二号”这样的任务表明，当下的成功取决于一个更为迫切的问题：能够承受太空严酷环境的硅材料。 

当前的差异化因素不在于芯片的生产地，而在于其设计是否足够坚固耐用。值得信赖、安全可靠且具备抗辐射能力的半导体解决方案——例如采用GF技术制造的解决方案——仍是任务就绪性的基石，确保每个系统在关键时刻都能无故障运行。 

通过融合先进制造技术、安全的全球供应链以及深度生态系统协作，GF正推动新一代航天创新。随着探索活动加速向地球轨道以外拓展，未来的发展道路将取决于支撑这些探索之旅的技术的韧性。 

阿富萨特·迪里斯在工业、政府和学术界拥有超过25年的从业经验。她的职业生涯涵盖数字电子和量子级联激光器设计，以及技术咨询、投资组合管理、STEM教育项目领导工作，以及技术销售和业务开发。 

她拥有普林斯顿大学电气工程博士学位、石溪大学电气工程硕士学位，以及纽约理工大学（现为纽约大学坦登工程学院）计算机工程学士学位。

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. 

通过 GF Accelerate，我们选择与经验丰富的风险投资领军者合作，而非单打独斗——以有限合伙人的身份参与美国、欧洲和新加坡的领先投资平台。 该生态系统涵盖Silicon Catalyst和Socratic Partners，以及欧洲首个专注于半导体、光子学和先进材料的基金Cloudberry；此外，最近加入的Playground Global其第四期基金（Fund IV）投资领域横跨计算、自动化和生命科学。每一项合作都深化了GF与致力于攻克AI数据中心及物理AI背后基础技术的创始人的联系。 

这些并非被动投资。它们使我们能够与创始人及投资者紧密合作，支持初创企业，并在研究、资本筹集与生产执行之间建立更紧密的反馈循环。 

将创新转化为可靠的硅片
对于初创企业及其投资者而言，通往硅片化的道路是风险最集中的环节，也是合作伙伴关系最为关键的环节。GF 直接与初创企业合作，帮助其完善技术并提供关键验证，在适当情况下利用更广泛的生态系统，并在合适的阶段支持其向量产过渡。 

我们提供关键资源支持，包括工艺设计套件（PDK）、多项目晶圆（MPW）计划、基础IP、参考设计、基础设施以及与GF技术专家的直接对接。这种方法有助于初创企业降低执行风险、加速学习进程，并建立对突破性创意可规模化应用的信心。 

展望未来
下一代计算时代的定义，不仅取决于渐进式进步，更取决于材料、架构和系统集成方面的根本性变革。初创企业将在这一颠覆性变革中发挥核心作用，尤其是在挑战超出任何单一组织能力范围的领域。 

GF致力于与那些同样秉持长期价值创造理念的创始人、投资者及科技领军人物携手共进。如果您正在开创未来，并寻求既能理解您的雄心壮志又能体谅您所面临限制的合作伙伴，我们坚信，最持久的成功故事正是通过携手共建而成就的。