Participated · Series C · Jun 2019

Fungible builds programmable data processing units (DPUs) designed to shift data centers from a compute-centric to a data-centric architecture. Its DPUs connect over standard IPs via Ethernet and local buses like PCI Express to CPUs, GPUs and storage, acting as a "super-charged data traffic controller" that offloads computations. The company says DPUs complement central and graphics processors and enable data centers to handle much larger volumes of data more efficiently, potentially unlocking new applications across industries such as self-driving, personalized healthcare, cloud gaming, finance, agriculture, call centers and manufacturing. Fungible plans to release timing and onboarding details for its DPUs later this year and says onboarding will not require customers to change existing applications, networking, or server design. Founded in 2015, the company has about 200 employees and will use the new capital to speed product development. The founders, CEO Pradeep Sindhu and Bertrand Serlet, have prior senior roles at Juniper Networks and Apple (and at Upthere), respectively. Fungible is developing a data center infrastructure solution designed to improve the economics, security and reliability of cloud data centers. The company offers a full-stack approach aimed at rethinking data-center architectures. Co-founded by Pradeep Sindhu (founder of Juniper Networks) and Bertrand Serlet (former Apple software engineer), Fungible is based in Santa Clara, CA. The company raised a Series A round to fund product development and near-term commercialization. Management says the funds will be used to continue developing its first phase of products, which were expected to roll out in 2018. Board and advisory changes accompanied the financing, reflecting investor involvement in company oversight.

Participated · Equity · Oct 2018

Berkeley Lights has developed a digital cell biology platform that combines bioscience, optofluidic light technology, and information to individually isolate, culture, assay and export thousands of live cells while providing visual and digital monitoring. The platform enables scientists to select and retrieve the best cells, accelerating drug discovery and development for large-molecule therapeutics. The company’s technology is used by biopharmaceutical companies working in large-molecule drug discovery and development. Berkeley Lights completed a $95M financing led by Nikon, with participation from existing investors Sequoia Capital, Walden-Riverwood Ventures, Black Diamond Ventures, and Paxion Capital, and additional new investors including Cota Capital, KTB Network, Atinum Investment, Shangbay Capital, AJS BioTree Healthcare Fund, and Varian. The company plans to use the funds to expedite solutions for cell therapy development and manufacturing, including CAR T and endogenous T-cells, and to increase research and development and expand commercial activities. Eric Hobbs, Ph.D., serves as CEO.

Participated · Series D · Apr 2018

Innovium develops the TERALYNX family of programmable switch silicon delivering 1Tbps to 25.6Tbps throughput with telemetry, low latency, programmability and large buffers, and a roadmap that scales to 51.2Tbps+. The company reports TERALYNX 7 in continued volume production, TERALYNX 5 ramping shipments for ToR, Edge and 5G designs, and the introduction of TERALYNX 8. Innovium says its products have been adopted and validated by market-leading OEM, cloud and ODM customers, including wins, deployments and trials at a majority of the top 25 cloud customers. Financially, Innovium reported revenue growth of over 5x in 1H 2020 versus 1H 2019 and achieved over 20% market share in 50G SerDes shipments. The business is headquartered in Silicon Valley, California. The company plans to use new capital to accelerate R&D innovation and expand customer engagements worldwide. Innovium develops the TERALYNX family of data-center optimized switching silicon, delivering products ranging from 3.2Tbps to 12.8Tbps with a focus on power efficiency, programmability, radix, buffers, and low latency. The company emphasizes a grounds-up platform designed to meet rising data-center demands from public and hybrid cloud, machine learning, analytics, storage, and video. Innovium plans to use new funding to drive production and volume ramps of its TERALYNX 12.8Tbps and 6.4Tbps switch family and to continue product roadmap innovation and market expansion. The article notes the TERALYNX silicon is seeing market traction and volume deployment at several industry-leading customers. Innovium is headquartered in Silicon Valley, California and highlights a team with experience delivering multiple generations of widely deployed data-center products. The company states its products are software-compatible across the TERALYNX family. Innovium develops high-performance TERALYNX switching silicon targeted at data center networks. Its TERALYNX family delivers software-compatible products ranging from 3.2Tbps to 12.8Tbps with claims of unmatched power efficiency, radix, programmability, buffers and low latency. The company says its data center–focused design and patented innovations enable it to leapfrog competing products and position it as a leader in programmable large-scale data center networks. Innovium plans to use the new funding to accelerate commercialization of the TERALYNX 12.8Tbps switch silicon family and to execute on its roadmap. The company is headquartered in Silicon Valley, California. As of this round Innovium has raised a total of $90 million in financing. Innovium develops networking infrastructure products aimed at next‑generation data centers and is based in San Jose, California. The company focuses on building hardware and software solutions for data center networking (as described in the article). In 2015 it completed two oversubscribed financing rounds that together raised over $50 million. The financings drew participation from strategic and financial investors, and led to multiple additions to Innovium’s board. Rajiv Khemani is the company’s CEO and founder. The article does not disclose operating metrics such as revenue or users.

Participated · Series A · Jan 2018

Eta Compute builds energy-efficient endpoint AI solutions, including the ECM3532 neural sensor processor and the TENSAI® Flow software and neural network compiler. The company’s patented Continuous Voltage and Frequency Scaling (CVFS) technology underpins its platform and targets multi-year battery life and energy-harvesting applications. Eta Compute offers board-level solutions and optimized neural networks for audio- and vision-based edge AI in smart home consumer IoT and industrial IoT markets. The company said it will expand from software into complete, commercially deployable system solutions and introduce new modules and software to accelerate design wins. Financially, Eta Compute closed a $12.5 million Series C, bringing total capital raised to $31.9 million. The company is based in Westlake Village, Calif. Eta Compute develops third-generation machine intelligence for mobile and edge devices, focusing on spiking neural networks (SNNs) that use biologically realistic neuron models. The company leverages its Delay Insensitive Asynchronous Logic (DIAL™) to power machine intelligence on constrained devices. Eta Compute aims to enable brain-inspired computation techniques and carry out machine intelligence based on these biologically inspired models. It completed an $8M Series A and said it will use the funds to accelerate its machine-intelligence hardware platform and AI applications. The company was founded in 2015 by Gopal Raghavan (CEO) and Dr. Narayan Srinivasa (CTO) and is based in Westlake Village, California.

Led · Series C · Dec 2017

Vayyar develops radar-imaging "radar-on-chip" sensor technology that uses MIMO antennas to deliver high-resolution mapping of surroundings. The company began by pursuing alternative breast-cancer screening and later expanded its technology into automotive, senior care, retail, smart home, commercial property and construction, offering products such as Vayyar Care (fall detection) and the Walabot handheld sensor. It has secured partnerships and customer relationships including Amazon (Alexa Together integration), Piaggio Group deployments, supply contracts with automakers in Japan and Vietnam, and a joint venture with Haier subsidiary HCH Ventures in China. Founded in 2011 by Miri Ratner, Naftali Chayat and Raviv Melamed, Vayyar plans to introduce a family of machine-learning-powered sensor solutions targeting robotics, retail, public safety and smart buildings. The company raised $108 million in a Series E, bringing total capital raised to over $300 million, and says the latest funding values it at over $1 billion. Vayyar engaged China International Capital Corporation as lead financial adviser to support investor outreach in China and has opened a new office there. Vayyar develops 4D radar-imaging chips, sensors and the software that interprets radar-generated images to detect and track people and objects while preserving privacy. Its technology is applied across automotive and IoT use cases; the company also produces a consumer product, the Walabot handheld sensor. Customers include automotive supplier Valeo and an unnamed "major Silicon Valley company" that is integrating Vayyar tech into smart-home products. The company primarily supplies technology to partners who productize it rather than focusing on direct-to-consumer sales. Vayyar says its software and algorithms differentiate it from other radar-imaging competitors. Management expects regulatory trends (for example, automotive safety requirements around detecting children left in cars) and growing privacy concerns to accelerate demand for its solutions. Vayyar Imaging develops 3D imaging sensors that create realtime, camera-free images capable of seeing through solid objects and mapping large areas. Its technology uses an embedded chip and advanced imaging algorithms to enable sensing in privacy-sensitive locations where optics cannot be used. The sensors are being applied across sectors including smart home, automotive, retail, robotics, medical, construction and agriculture. Vayyar says its mission is to improve health, safety and quality of life worldwide, and that demand from diverse industries requires rapid scaling. Following a $45 million Series C, total capital raised to date is $79 million; the company will use proceeds to expand into new industries, grow its global team and diversify its sensing product offerings. Vayyar also markets a consumer handheld device (Walabot DIY) and planned demonstrations of its latest automotive and smart home sensors at CES 2018. Vayyar builds compact sensors that use radio waves to create 3D images of obstructed objects — from root systems and piping to tissue behind skin. The company works with chip manufacturers to deliver a reference design, tailored antennae and an API so customers can plug the sensor into phones, devices or ground‑installed instruments. Vayyar launched a board for external tinkering to accelerate use‑case discovery and is starting to roll out products more broadly as augmented reality interest grows. Targeted applications mentioned include low‑cost breast‑cancer detection, milk fat measurement during production, irrigation profiling and infrastructure inspection. The startup faces competition from ultrasound, MRI and other chip/sensor makers but positions its tech as lower‑cost and already in use by a number of companies. Financially, the company has raised $22M in the latest venture financing and has raised $34M to date.