Participated · Equity · Jul 2026

Xsight Labs develops programmable, high-performance networking silicon, notably the E1 Data Processing Unit (DPU) and X2 Ethernet switch. The E1 is the company's first shipping 800G DPU and has completed SONiC-DASH Hero 800G validation, demonstrating sustained throughput of more than 14 million connections per second, hardware-accelerated inline crypto, and an ultra-efficient sub-75W power profile. The X2 is a fully software-defined 12.8 Tbps Ethernet switch with a reprogrammable data plane via XISA, deterministic sub-700ns latency, and an under-200W power envelope. Xsight emphasizes a fully open, standard Linux environment (native Linux, DPDK, SONiC) and positions its products for hyperscalers, service providers, OEMs, and constrained environments like satellite communications. Founded in 2017 and headquartered in Tel Aviv with additional offices in Kiryat Gat, Haifa, Boston, Raleigh, San Jose, and Yerevan, the company has reported multiple design wins and evaluations by Tier-1 customers. Financially, Xsight recently raised more than $300 million at a $2.8 billion post-money valuation to fund product roadmap expansion, team growth, and scaled manufacturing and delivery.

Led · Equity · Aug 2025

SiMa.ai builds the Modalix MLSoC platform and associated System-on-Modules to enable real-time Physical AI at the edge, targeting robotics, autonomous systems, and industrial automation. Its platform emphasizes a sophisticated memory hierarchy and power-efficient compute to support complex workloads such as large language models and vision-language models on edge devices. SiMa.ai has integrated Micron’s LPDDR5X into Modalix to increase bandwidth efficiency and power optimization. The company pairs its hardware with Palette software to simplify deployment and to provide a path from prototype to production. SiMa.ai positions Modalix as a validated Physical AI solution within an ecosystem that includes partners such as Arm, Cerence AI, L&T Technology Services, Synopsys, TSMC, and Wind River. The Micron investment is intended to deepen hardware/software co-optimization and leverage Micron’s supply-chain relationships to accelerate customer deployments.

Participated · Seed · Feb 2025

Founded in 2024 and based in Portland, OR, AheadComputing is developing a breakthrough CPU microarchitecture that marries RISC-V extensibility with proprietary design to maximize per-core performance for AI inference, agentic AI, and other data-center tasks. The 120-person team, which collectively holds hundreds of compute patents and has shipped more than 70 products in prior roles, is currently building its first high-performance general-purpose CPU. The company positions its architecture as a critical complement to GPUs, arguing that CPU performance is still essential as 70% of data-center workloads are expected to be AI-related by 2030. Proceeds from recent funding are earmarked for R&D, software innovation, and test-chip development, as well as continued team expansion. Key technical collaborations include Alchip, Cadence, Skyechip, and Tenstorrent. To date, the company has raised $53 million, giving it runway to bring its initial product to market and further integrate RISC-V into mainstream AI infrastructure.

Participated · Series B · Oct 2024

Terray Therapeutics develops small‑molecule therapeutics by combining generative AI, ultra‑high‑throughput experimentation, biology, medicinal chemistry, automation, and nanotechnology. Its integrated AI platform, tNova, and chemistry foundation model COATI power both internal programs and partnered discovery efforts. The company reports it has quantitatively measured more than 5 billion target–ligand interactions in the past three years—roughly 50× the entirety of publicly available chemistry data—and that this dataset is doubling annually. Terray applies these capabilities to an internal pipeline focused on immunological diseases and to multi‑target partnerships with Bristol Myers Squibb and Calico. Proceeds from the Series B will advance internal programs into the clinic and expand the tNova platform and associated AI tools. Since its Series A in 2021, Terray has raised over $200 million in equity financing to date. Terray Therapeutics, led by CEO Jacob Berlin, develops small-molecule therapeutics by integrating generative AI with high-throughput experimentation, biology, medicinal chemistry, automation, and nanotechnology. Its platform combines experimentation and computation to build ligand- and structure-based models aimed at accelerating discovery of challenging therapeutic targets. The company plans to develop comprehensive chemistry foundation models and will leverage NVIDIA DGX Cloud and the NVIDIA BioNeMo cloud service to scale and optimize those models. Terray intends to use its proprietary experimental datasets to train both ligand-based and structure-based generative models to propel its pipeline toward the clinic. Financially, Terray received an investment from NVentures alongside a syndicate of new and existing institutional investors; the amount of the financing was not disclosed. Named individual investors in the syndicate include John Maraganore and Bassil Dahiyat. Terray Therapeutics builds integrated computational and experimental platforms that generate scaled, agnostic chemical data to power computational learning and reveal new molecular interactions. The company combines AI, synthetic chemistry, automation and nanotechnology to propel drug discovery into the information age. Terray is led by CEO Jacob Berlin and is advancing a preclinical pipeline alongside discovery partnerships. The company intends to use the Series A proceeds to advance its preclinical programs and its discovery collaborations. This Series A follows a previously unannounced $20 million seed round co-led by Digitalis Ventures and Two Sigma Ventures. Terray is headquartered in Pasadena, California.

Participated · Safe · Oct 2024

TensorWave operates an all-AMD AI cloud focused on high-performance, memory-intensive workloads, delivering high-bandwidth, memory-optimized infrastructure powered by AMD Instinct series GPUs. The company currently runs one of the largest AMD-based AI training clusters in North America with 8,192 AMD Instinct MI325X GPUs online. TensorWave has secured more than 2 gigawatts of long-term data center capacity to support growing adoption from enterprise, research, and AI-native customers. Following a Series A in May 2025, TensorWave completed a $350 million Series B at a $1.55 billion valuation to fund global expansion and next-generation MI355X GPU deployments. The company is investing in its Las Vegas headquarters and plans to expand hiring across engineering, infrastructure, operations, sales, and customer success. Customers mentioned include next-generation AI companies such as Fireworks AI and Luma AI, which are using TensorWave’s AMD-based infrastructure for large-scale generative AI workloads and production inference.