Heng (Henry) Huang

Hark is an artificial intelligence lab building the most advanced personal intelligence in the world. Were pairing our own foundation models with bespoke hardware to create a universal interface between people and machines. Hark has raised over $700 million in Series A funding. With these resources, were hiring some of the best talent from across the industry.

Axiado is a leading hardware-anchored platform security solutions company deploying a novel, AI-driven approach to platform security against ransomware, supply chain, side-channel, and other cyberattacks in the growing ecosystem of cloud data centers, 5G networks, and other disaggregated compute networks. We offer a new processor called the Trusted Control/Compute Unit (TCU) that redefines security from the ground-up. Its hardware-anchored and AI-driven security technologies include Secure Vault root-of-trust/cryptography core and per-platform Secure AI pre-emptive threat detection engine.

At Multiscale AI, we’re revolutionizing semiconductor manufacturing with advanced, agentic AI and digital twins. We partner with leading semiconductor manufacturers to tackle their most complex Semiconductor Manufacturing Technology Optimization (SMTO) challenges. Our platform combines Bayesian digital twins, purpose-built ML models, and AI agents to extract actionable insights from limited silicon data and physics simulations. By unifying fragmented data, delivering domain-optimized ML, and enabling scalable, fab-ready applications, we help customers reduce costly silicon trials, speed up innovation cycles, and optimize fab processes at scale. Our team brings together experienced scientists, AI/ML engineers, software developers, and materials experts, united by a commitment to build comprehensive, reliable, and scalable solutions that make semiconductor manufacturing more efficient, intelligent, and competitive.

Oxide is building a new kind of server. True rack-scale design, built with the innovations of cloud hyperscale technology, to make running on-premises compute infrastructure as easy as cloud.

Semiconductor interconnects

ChipAgents is an agentic AI company building AI-native chip design and verification workflows. Its platform transforms design intent and specifications into production-ready RTL, verification assets, and automated root-cause analysis, delivering 10-100× productivity gains across production semiconductor development workflows. The platform is proven by IC designers, verification engineers, architects, and embedded systems teams at more than 120 semiconductor companies worldwide. Founded in 2024 and headquartered in Santa Clara, California, ChipAgents is at the forefront of AI innovation in semiconductor design.

We are nEye.ai, an optical switch startup founded in 2020 to revolutionize the future of data centers. nEye’s MEMS-based silicon photonics optical circuit switches (OCS) eliminate critical bottlenecks in AI processing by enabling direct optical connections among thousands of GPUs and memory units. The companys SuperSwitch is an ultra-low power consumption, high radix, compact chip-scale design, offering hyperscale data centers enhanced performance, efficiency, and scalability. www.neye.ai.

E-Space is a global space company focused on bridging Earth and space with the most sustainable low earth orbit (LEO) network that is expected to reach over one hundred thousand multi-application communication satellites to help businesses and governments securely and affordably access the power of space to solve problems on Earth. Founded by industry pioneer Greg Wyler, E-Space is focused on democratizing space and transforming industries by bringing down the cost of space-based communications, raising the level of satellite system resiliency and setting a new standard in sustainable space infrastructure that will effectively minimize and reduce space debris and destruction while preserving access to space for future generations.

Focused on making AI infrastructure radically more efficient and sustainably scalable, by building the worlds densest AI compute platform, integrating innovations across technology, chip and system architecture, software, and datacenter infrastructure

Nubis is developing a breakthrough optical technology platform that enables the industry’s highest density and lowest power optical interconnect. The first product on this platform, the XT1600™optical engine, has 1.6 Tbps full-duplex capacity optimized for unretimed direct-drive operation, connecting directly with host ASIC SerDes designed for copper links. The highly integrated optical engine has 16 transmit and 16 receive fibers of 100G each, which can be configured to support various aggregate system port speeds including standardized Ethernet 400G-DR4 or 800G-DR8.

Industrial humanoids hard at work.

Silicon Box is an advanced semiconductor packaging company, specializing in cutting-edge integration technology and manufacturing processes. We offer solutions that enable chiplet architecture, as well as high performance alternatives to traditional packaging schemes. Leveraging our proprietary technology, 30 years of multi-sectoral expertise, and relationships with best-in-class partners, we strive to solve the unique challenges of chiplet adoption in order to build the emergent technologies shaping the world around us today. Silicon Box was founded in 2021 by Dr. Byung Joon (BJ) Han, Dr. Sehat Sutardja and Weili Dai. Dr. Han is the inventor of many of the most advanced packaging solutions in today’s market, and his team at Silicon Box holds standing records for yield in advanced packaging technology at wafer level production. Dr. Sutardja introduced the concept of chiplets at the International Solid-State Circuits Conference (ISSCC) in 2015. He and Weili Dai founded Marvell Technology Group (NASDAQ: MRVL) in 1995, and ran the company for twenty years.

Metalenz, founded in 2016 is the first company to commercialize meta-optics. The company is backed by leading investors including 3M, Applied Ventures LLC, Intel Capital, M Ventures and TDK Ventures. The company has the exclusive worldwide license to the portfolio of foundational intellectual property relating to metasurfaces developed in the Capasso Lab at Harvard University and has more than 20 patents on innovations that simplify and improve optical devices across multiple markets

Inference is the engine that powers AI, and Groq was built from the silicon up to deliver the worlds fastest inference at scale. We pioneered the LPU—the first processor designed specifically for AI inference—and are transforming that innovation into a global cloud platform powering production AI workloads. With the capital, infrastructure, and team to execute, were uniquely positioned to define the next era of AI infrastructure. The opportunity is massive, and its still wide open. Now lets go build it!

Analog Inference is revolutionizing neural computing. Founded by veterans of the computing industry and backed by Khosla ventures. The company is currently developing its first generation of products. Investors and advisors also include Andy Bechtolsheim (co-founder of Sun Microsystems), Pradeep Sindhu (co-founder of Juniper Networks), and Atiq Raza (CEO of Raza Microelectronics), and Ajit Medhekar (Entrepreneur and Angel Investor).

Hark is an artificial intelligence lab building the most advanced personal intelligence in the world. Were pairing our own foundation models with bespoke hardware to create a universal interface between people and machines. Hark has raised over $700 million in Series A funding. With these resources, were hiring some of the best talent from across the industry.

Axiado is a leading hardware-anchored platform security solutions company deploying a novel, AI-driven approach to platform security against ransomware, supply chain, side-channel, and other cyberattacks in the growing ecosystem of cloud data centers, 5G networks, and other disaggregated compute networks. We offer a new processor called the Trusted Control/Compute Unit (TCU) that redefines security from the ground-up. Its hardware-anchored and AI-driven security technologies include Secure Vault root-of-trust/cryptography core and per-platform Secure AI pre-emptive threat detection engine.

At Multiscale AI, we’re revolutionizing semiconductor manufacturing with advanced, agentic AI and digital twins. We partner with leading semiconductor manufacturers to tackle their most complex Semiconductor Manufacturing Technology Optimization (SMTO) challenges. Our platform combines Bayesian digital twins, purpose-built ML models, and AI agents to extract actionable insights from limited silicon data and physics simulations. By unifying fragmented data, delivering domain-optimized ML, and enabling scalable, fab-ready applications, we help customers reduce costly silicon trials, speed up innovation cycles, and optimize fab processes at scale. Our team brings together experienced scientists, AI/ML engineers, software developers, and materials experts, united by a commitment to build comprehensive, reliable, and scalable solutions that make semiconductor manufacturing more efficient, intelligent, and competitive.

Oxide is building a new kind of server. True rack-scale design, built with the innovations of cloud hyperscale technology, to make running on-premises compute infrastructure as easy as cloud.

Semiconductor interconnects