Absci builds a generative-AI drug-creation platform that designs and optimizes biologics, including antibodies, through in-silico modeling tightly coupled with wet-lab validation. Its platform focuses on creating better biologics faster by using AI to explore novel biology, run de-risked therapeutic programs, and power both internal and partnered pipelines.
Axiado builds AI-driven platform security and system management solutions for accelerated computing and AI data-center infrastructure. Its technology combines silicon-level root of trust, cryptography, authentication, attestation, and preemptive threat detection to help protect servers and infrastructure from network, peripheral, and physical side-channel attacks.
Ayar Labs builds specialized AI infrastructure leveraging photonics and custom ASIC-based solutions to enable efficient, high-bandwidth, low-latency data center interconnects and AI compute scaling. Its platform focuses on delivering purpose-built multi-chip packages and photonic interconnects to power next-generation AI and hyperscale computing efficiently.
Bodo.ai builds a next-generation compute engine and auto-parallelizing JIT compiler that converts standard Python (Pandas, NumPy, and more) into supercomputing-style parallel execution. Its platform focuses on accelerating large-scale data processing and AI workloads while cutting cloud and energy costs, without requiring users to rewrite their code.
Celestial AI builds cutting-edge AI infrastructure centered on optical interconnect technology that provides high-bandwidth, low-latency networking and scale-up compute connectivity for AI data centers. Its platform focuses on enabling large-scale AI model training and inference with performance and scalability by rethinking connectivity, switching, and packaging at the silicon and network level.
Cellular Intelligence builds foundation models for human stem cells and the “digital twins” of embryonic development to produce human tissue for cell therapies at scale. Its platform focuses on making cell replacement therapies faster, more predictable, and more affordable by using AI-driven models to design, optimize, and manufacture next-generation cell therapies.
CHAI AI builds conversational AI systems that let users create and interact with highly engaging, customizable chatbots. Its platform focuses on delivering rich, emotionally resonant experiences by tuning models for depth of interaction, user preference alignment, and overall engagement quality.
Chipletz builds Smart Substrate™ advanced packaging that lets customers integrate heterogeneous chiplets—compute, I/O, security, HBM, and more—into a single high-density system without an interposer. Its platform focuses on reimagining AI and HPC system architecture by using chiplet integration to boost performance, flexibility, and bandwidth beyond traditional monolithic SoCs.
Classiq builds a quantum software development platform that lets developers model, synthesize, and compile complex quantum algorithms from high-level functional descriptions instead of hand-coding gate-level circuits. Its platform focuses on generating hardware-aware, optimized quantum circuits at scale — making it easier for enterprises and researchers to design, analyze, and deploy quantum applications across diverse cloud and quantum hardware backends.
Cohere builds large-language and natural-language processing models that enable developers and businesses to integrate powerful AI tools for text generation, analysis, search, and content moderation. Its platform prioritizes enterprise-readiness by offering secure, scalable, and customizable AI solutions that help teams unlock insights, improve productivity, and automate complex language tasks.
DriveNets builds cloud-native networking software that disaggregates traditional routing infrastructure into scalable, software-based network clouds running on open white-box hardware. Its Network Cloud and Network Cloud-AI platforms focus on helping service providers, cloud operators, and AI infrastructure builders scale core routing and Ethernet-based AI fabrics with greater flexibility, automation, and cost efficiency.
Efinix builds compact, power-efficient FPGA families such as Trion™ and Titanium™ that bring high-performance programmable logic to edge, embedded, and AI applications. Its devices focus on enabling AI acceleration, real-time analytics, and custom digital logic in space- and power-constrained systems like machine vision, automotive, and handheld medical devices.
Eliyan builds high-speed die-to-die and chip-to-chip interconnect IP and chiplets that enable dense, low-power links across both advanced and standard packaging technologies. Its platform focuses on breaking IO and memory bottlenecks for AI and HPC by delivering interposer-class bandwidth, low latency, and flexible multi-die integration without requiring costly exotic packaging.
Essential AI builds open-platform deep learning systems that aim to accelerate foundational research and make advanced AI tools broadly accessible. Its platform focuses on advancing frontier model capabilities and enabling developers and researchers to experiment, build, and innovate with cutting-edge AI systems.
Ethernovia builds automotive-grade Ethernet PHYs and network-switching solutions that power next-generation software-defined vehicles. Its platform focuses on delivering high-bandwidth, low-latency and energy-efficient data transport to support advanced driver assistance, autonomous driving, and rich in-vehicle sensor workloads.
Featherless AI builds a serverless inference platform that gives developers and enterprises API access to a broad catalog of open-weight AI models. Its platform focuses on simplifying model deployment, expanding access to specialized open models, and reducing the operational complexity of running inference infrastructure at scale.
Fireworks AI builds cloud-native AI infrastructure that enables developers and organizations to deploy and scale generative and inference workloads without managing GPU infrastructure directly. Its platform focuses on simplifying model deployment and offering serverless, flexible options so users can run production-scale AI efficiently.
Generative Bionics builds intelligent humanoid robot companions, combining advanced robotics, artificial intelligence, and design to operate safely and reliably in real-world environments. Its platform focuses on amplifying human potential by deploying human-centric humanoids across manufacturing, healthcare, and in-home assistance as it scales toward industrial production.
Genki Robotics builds humanoid robotics technology aimed at advancing general-purpose robotic capabilities. Its work focuses on developing and prototyping the core systems required to enable capable, adaptable humanoid robots as advances in AI unlock new possibilities for real-world deployment.
Hark builds personal AI systems designed to create a new universal interface between humans and machines. Its platform combines foundation models, agentic multimodal software, persistent memory, and AI-native hardware to help users offload digital tasks and interact with technology through more proactive, personalized, and human-centered intelligence.
Hugging Face builds an open ecosystem for machine learning models and tools that enable practitioners to share, deploy, and collaborate on AI models across tasks like natural language processing, computer vision, and more. Its platform emphasizes openness, community-driven development, and accessibility — making state-of-the-art AI tools broadly available for developers and researchers.
Kepler Computing builds distributed-compute technology designed to harness many interconnected nodes as a single, more powerful system. Its approach focuses on extending the effective life of Moore’s law by orchestrating clusters of commodity hardware into resilient, scalable compute fabrics for data- and simulation-heavy workloads.
Liquid AI builds efficient, full-stack AI solutions that enable deployment of foundation models across devices — from edge hardware to cloud — by optimizing for compute efficiency and flexibility. Its platform focuses on offering customizable, high-performance AI that delivers low-latency inference and privacy-aware functionality directly on-device.
Luma AI builds multimodal generative AI systems that let users create high-fidelity images, videos, and 3D content from text, images, or video inputs. Its platform focuses on empowering creators — from filmmakers to designers — with tools that turn ideas into photorealistic visuals through intuitive, AI-driven workflows.
Moreh builds AI infrastructure software that turns large clusters of heterogeneous accelerators into a single, efficient “AI computer” for frontier-scale models. Its platform focuses on solving hyperscale challenges like parallelization, scalability, and performance optimization while avoiding vendor lock-in for data-center operators.
Odyssey builds general world models that help AI systems understand, simulate, and interact with the physical world across complex, dynamic environments. Its platform focuses on advancing real-time, physically accurate world simulation for applications across robotics, gaming, autonomous systems, and interactive software.
Pliops builds the Extreme Data Processor (XDP), a storage and data-acceleration engine that offloads and optimizes database, analytics, and AI data-path operations. Its solution focuses on delivering up to order-of-magnitude speed-ups while cutting infrastructure costs by reshaping how data is stored, indexed, and accessed in flash-based systems.
RadixArk builds frontier AI infrastructure for large-scale inference and training, with a focus on making advanced model-serving systems more open, efficient, and accessible. Its work centers on SGLang and related infrastructure that help developers reduce inference cost, improve throughput, and support production-scale AI workloads across diverse hardware environments.
Ranovus builds co-packaged optics and silicon-photonics optical engines, such as its ODIN® platform, to deliver multi-terabit interconnects for next-generation data centers. Its solutions focus on providing scalable, power-efficient bandwidth for AI/ML-optimized infrastructure where traditional electrical links hit capacity and energy limits.
Recursive builds AI research infrastructure focused on recursive self-improvement, open-ended discovery, and automated scientific experimentation. Its technology aims to create AI systems that can conduct experiments to safely improve themselves over time, starting with AI research and model development before expanding to broader scientific domains.
Runway builds generative AI models and creative tools that let users turn text, images, and reference media into high-fidelity videos and other rich multimedia content. Its platform focuses on giving creators and enterprises intuitive, production-ready workflows for cinematic video generation, editing, and world simulation powered by frontier video and multimodal models.
Scale AI builds infrastructure and data-labeling services that help companies prepare high-quality training data and build reliable, scalable AI systems. Its platform focuses on enabling rapid, large-scale AI development by streamlining data annotation, model evaluation, and pipeline orchestration.
ScaleFlux builds NVMe SSDs and CXL-enabled memory products with on-device computational storage that compress, process, and manage data directly on the drive. Its platform focuses on helping data-center and AI workloads keep up with data growth by boosting performance, density, and endurance while simplifying infrastructure and lowering TCO.
Sciforium builds byte-native multimodal foundation models alongside a proprietary, high-efficiency serving platform for real-time AI applications. Its platform focuses on rethinking AI infrastructure so developers can run powerful, multimodal models at a fraction of traditional inference cost, backed by deep hardware collaboration with partners like AMD.
Seekr builds AI-powered decision-support systems for government and critical-infrastructure environments where accuracy, explainability, and trust are paramount. Its platform focuses on delivering AI solutions that help leaders make high-stakes decisions with clarity, reliability, and full accountability.
SiFive builds customizable RISC-V processor IP portfolios that span low-power embedded cores to highly scalable vector and matrix engines for AI and data-center workloads. Its offerings focus on giving chip designers flexible, high-performance CPU IP so they can differentiate silicon across edge, automotive, and datacenter AI systems on an open ISA.
Tensormesh builds AI inference optimization software that reduces the cost and latency of serving large language models. Its caching-accelerated platform, built around distributed KV-cache reuse and LMCache technology, focuses on eliminating redundant computation so enterprises can improve GPU utilization while keeping control of their data and deployment environments.
TensorWave builds high-performance AI cloud and HPC infrastructure for training and inference workloads. Its platform is built exclusively on AMD Instinct™ accelerators, offering scalable, memory-optimized compute through large AMD-native GPU clusters, liquid cooling, and optimized networking designed for demanding AI workloads.
Teramount builds fiber-to-chip connectivity solutions using its Universal Photonic Coupler to make attaching optical fibers to silicon-photonics chips far more tolerant and scalable. Its technology focuses on simplifying high-volume optical packaging so data-center and telecom systems can unlock higher-speed, lower-cost photonic interconnects.
Thinking Machines builds multimodal AI systems and tools that let organizations fine-tune and customize frontier models for their own needs. Its platform focuses on making advanced AI more understandable, consistent, and broadly useful through open science, human-AI collaboration, and highly configurable model behavior.
Uniphore builds a comprehensive “Business AI Cloud” that unifies data, knowledge, agents, and models so enterprises can deploy AI-driven workflows for customer service, sales, HR, and more. Its platform focuses on delivering secure, scalable, and enterprise-ready AI applications that embed directly into organizational processes.
Upstage builds enterprise-grade large language models and document-processing engines that help businesses automate workflows and improve operational efficiency across sectors like healthcare, finance, insurance, and manufacturing. Its platform focuses on transforming how teams handle documents and information, using AI to streamline labor-intensive tasks.
Video Rebirth builds AI-native video generation technology that bridges creative imagination and production-ready visual content. Its BACH industrial-grade AI engine focuses on generating coherent, controllable, physically consistent 30fps video assets so creators can transform ideas into commercial-ready visuals and interactive worlds with fewer technical barriers.
Vultr builds cloud infrastructure and scalable compute services that allow developers, startups, and businesses to deploy applications — including AI workloads — on demand, with flexible resource provisioning. Its platform focuses on providing reliable, affordable, and easy-to-use infrastructure to lower the barrier for running cloud-based services.
Wayve develops embodied AI for autonomous driving, using end-to-end learning to help vehicles perceive, reason, and navigate complex real-world environments. Its AV2.0 approach focuses on training foundation models for autonomy on large-scale driving data so automakers and mobility platforms can deploy adaptable, production-ready self-driving capabilities across global markets.
World Labs builds spatial-intelligence AI systems that enable creation and interaction with detailed, navigable 3D environments. Its platform focuses on giving AI “world model” capabilities — letting models perceive, generate, reason about, and interact with the 3D world to unlock new forms of design, simulation, immersive storytelling, and embodied intelligence.
Xanadu builds photonic quantum computing systems designed to scale toward fault-tolerant quantum computation using light-based qubits. Its platform focuses on combining quantum hardware, photonic chip manufacturing, and software tools to make practical, large-scale quantum computing more manufacturable and accessible for advanced scientific and industrial workloads.
Xelera builds the Xelera Suite, a hardware-independent acceleration software platform that boosts data-center and cloud workloads by bridging applications to GPUs, FPGAs, and other accelerators. Its software focuses on delivering ultra-low-latency, high-throughput computing and TCO savings by making accelerator adoption plug-and-play across cloud, financial, and real-time analytics environments.
xAI builds large-scale AI and machine learning technologies intended to push forward model capabilities and research in next-generation AI. Its platform emphasizes high-performance models aimed at powering advanced applications and expanding what’s feasible in AI-driven tasks.
Xsight Labs builds high-performance connectivity silicon as a fabless semiconductor company focused on end-to-end Ethernet and edge-processing solutions for AI and cloud infrastructure. Its products focus on delivering high-bandwidth, low-latency links that re-architect the foundation of cloud, AI factory, and edge compute networks.
Zyphra designs multimodal AI architectures optimized for efficient agentic behavior across edge and on-prem environments. Its Zamba and ZAYA hybrid model families achieve frontier-class reasoning with substantially reduced computational overhead, enabled by training on a fully AMD-native hardware and software platform.
