Startup Funding: September 2023

Chip-to-chip and data center I/O drew investor interest in September, including support for several startups developing Compute Express Link (CXL) solutions. Elsewhere in the data center, several large rounds went to companies developing AI accelerators. And at the edge, startups are building unique ways to handle AI at very little power by initially processing data directly at the sensor.

On the manufacturing and test side, one startup is splitting nanostructures in half using atomic layer etching, while others are introducing e-beam inspection and scanning probe microscopy metrology equipment.

Here are some highlights from September funding rounds [find the full sortable table here].

Chips

Enfabrica raised $125.0M in Series B funding led by Atreides Management, joined by Sutter Hill Ventures, Nvidia, IAG Capital Partners, Liberty Global Ventures, Valor Equity Partners, Infinitum Partners, and Alumni Ventures. Enfabrica designs what it calls Accelerated Compute Fabric (ACF) I/O chips for scalable multi-terabit-per-second data movement between GPUs, CPUs, accelerators, memory, and networking devices in data centers. Its Accelerated Compute Fabric Switch (ACF-S) devices enable switching and bridging between heterogeneous compute and memory resources in a single silicon die using standards-based interfaces including multi-port 800 Gigabit Ethernet networking and high-radix PCIe Gen5 and CXL 2.0+. It combines CXL.mem disaggregation, performance/capacity layering, and Remote Direct Memory Access (RDMA) networking to implement a high-bandwidth, high-capacity, latency bound memory hierarchy for large-scale AI compute, which the startup claims reduces the number of GPUs and CPUs required for a task. Founded in 2020, it is based in Mountain View, California, USA.

Kandou raised $72.3M in the first close of an equity financing round that is expected to bring in $100.0M. Kandou provides high-speed chip-to-chip link solutions. The company is preparing to launch a multiprotocol retimer for data centers to support PCIe 5.0 and CXL 2.0. It also offers a USB-C multiprotocol retimer solution with USB4 support for consumer and networking applications and SerDes PHY IP using its Chord signaling for ultra-short connections between chiplets and in MCMs. Funds will be used for global expansion. Founded in 2011, it is based in Saint-Sulpice, Switzerland, and has raised $280M to date.

Panmnesia raised $12.5M in seed funding led by Daekyo Investment and joined by SL Investment, Smilegate Investment, GNTech Venture Capital, Time Works Investment, Yuanta Investment, and Quantum Ventures Korea. Panmnesia develops CXL semiconductor IP and data processing acceleration hardware and software solutions, including for CXL 3.0. Among its products is an SoC that offers CPU-to-device, CPU-to-memory, and device-to-memory connections integrated with near-data processing units (DPU) and processing-in-memory (PIM) units. A spin out from KAIST founded in 2022, it is based in Seongnam, South Korea.

AI hardware

d-Matrix raised $110.0M in Series B funding led by Temasek, joined by Playground Global, Microsoft’s M12, Nautilus Venture Partners, Entrada Ventures, Industry Ventures, Ericsson Ventures, Marlan Holdings, Mirae Asset, Cortes Capital, Archerman Capital, TGC Square, Lam Capital, and Samsung Ventures. d-Matrix has developed a digital in-memory computing (DIMC) architecture that targets generative AI and large language model (LLM) data center inference workloads, along with accompanying ML tools, software, and algorithms. It currently offers several chiplets based on the DIMC technology, with its latest using the Bunch of Wires PHY interconnect. Funds will be used for commercialization of its next-generation chiplet-based inference compute platform. Founded in 2019, it is based in Santa Clara, California, USA.

Kneron added $49.0M to its Series B round from investors including Foxconn Co-GP Fund and Alltek, bringing the round to $97.0M. Kneron provides AI SoC processors based on the company’s reconfigurable NPU architecture and integrated software solutions for on-device edge AI inferencing. It supports both CNNs and transformers. Kneron’s latest chip is an auto-grade NPU that supports lightweight GPT LLMs and applies transformers to image-based applications to process time-series and consider image data holistically, an approach the company claims improves the accuracy of image-based applications by at least 30%. Funds will be used to accelerate the deployment of advanced AI, focusing on nano GPT solutions for automotive. Founded 2015, it is based in San Diego, California, USA.

Upmem received €7.0M (~$7.5M) in a combination of equity from the European Innovation Council Fund, Partech, Western Digital Capital, and C4 Ventures and a grant from the EIC. Upmem is developing processing-in-memory accelerators for AI and big data applications. The startup says its processor architecture allows data-intensive operations to be computed with hundreds of programmable DPU coprocessors integrated directly into DRAM, making it possible to modify large data analytics and AI applications to run on the coprocessors using high-level programming languages. Founded in 2015, it is based in Grenoble, France.

Aspinity drew $5.0M in Series B funding from current investors Anzu Partners, Birchmere Ventures, Mountain State Capital, Riverfront Ventures, and new investor Unitrontech. Aspinity designed an analog machine learning core that it says enables always-on sensing solutions at near-zero power that detect and inference from raw, analog sensor data. To lower power, it keeps the MCU and other digital processors asleep until meaningful data are detected. As part of the deal, Aspinity and Unitrontech plan to work together on aftermarket and integrated AI solutions for automotive applications. Founded in 2015, it is based in Pittsburgh, Pennsylvania, USA, and has raised over $19M to date.

Oculi received a $1.0M investment from the Luminate NY accelerator program. Oculi has developed a single chip software-defined vision sensor for real-time vision intelligence at the edge. Its technology combines sensing and in-pixel digital processing to lower latency to under a microsecond and reduce bandwidth and external post-processing by only transmitting actionable data rather than the entire scene. Real-time optimization via software allows the same hardware product to be used for multiple markets. Oculi says its neuromorphic sensor mimics the visual system in selectivity, parallel processing, and efficiency. Founded in 2019, it is currently based in Baltimore, Maryland, USA, but plans to move its headquarters to Rochester, New York.

EDA

Machine Discovery raised £4.5M (~$5.5M) in funding led by BGF and East Innovate, joined by Foresight WAE Technology, UK Innovation and Science Seed Fund, and Oxford Technology Management. Machine Discovery uses machine learning to accelerate compute-intensive simulation and optimization tasks. Its initial focus is analog IC design, where it works as a companion to existing tools and simulators. The platform creates neural network-based models from conventional simulation outputs for real-time prediction, then uses optimization and sampling AI algorithms to combine simulation and neural network model outputs for faster design space exploration. It also has applications in fusion power plant design and other simulation-intensive research. Funds will be used for hiring. A spin out from the University of Oxford founded in 2019, it is based in Oxford, UK.

Manufacturing & equipment

AlixLabs raised SEK 40.0M (~$3.6M) in funding from Navigare Ventures, Industrifonden, and FORWARD.one. AlixLabs develops Atomic Layer Etch Pitch Splitting technology (APS) for manufacturing of nanostructures. The technique uses atomic layer etching to split nanostructures in half with a single process step by using walls of inclined structures, fabricated with dry etching or epitaxial growth, as a mask. APS makes it possible to split a 40nm-wide feature into two features of 10nm at half the pitch, which can form the basis of transistors or electrical contacts and metal interconnects. AlixLabs says the process is scalable to smaller structures below 10nm and works with single-exposure immersion lithography and extreme ultraviolet (EUV) lithography as well as multiple patterning techniques. “This investment marks a significant milestone as we work towards our goal: to have our meticulously developed process adopted by circuit manufacturers in their most advanced production by 2025,” said Jonas Sundqvist, CEO of AlixLabs. Based in Lund, Sweden, it was founded in 2019.

Test, measurement & inspection

Nearfield Instruments added €16.0M (~$17.1M) to its latest investment round that includes ING Corporate Investments, Invest-NL, Innovation Industries, and TNO, bringing the round to €27.0M. Nearfield Instruments supplies a semiconductor scanning probe microscopy (SPM) system that can measure 3D surfaces. The company says its Feedforward Trajectory Planner imaging mode can be used for non-destructive in-line process monitoring of 3nm and below gate-all-around FETs (GAA FETs), while a parallel architecture of miniature SPMs increases throughput. Funds will be used to diversify its product portfolio and accelerate mass production, including bringing a non-destructive subsurface nanoscale metrology product to market. Founded in 2016 as a spin off from research organization TNO, it is based in Rotterdam, the Netherlands.

Photo electron Soul raised JPY 730.0M (~$5.0M) in financing led by Ushio. PeS provides photocathode e-beam systems for semiconductor inspection. The system allows for non-contact testing of transistors by selective radiation of an e-beam, along with detection of particles hidden at the bottom of a deep hole with the aspect ratio of 30 or more by instant control of the beam dose. PeS has also developed a technology it calls Selective e-Beaming (SeB) to enable e-beaming of samples with the desired probe current and location at the single-pixel level. The company claims its system can improve inspection throughput by 10x compared to current technology. Founded in 2015 based on R&D from Nagoya University, it is based in Nagoya, Japan.

UnitX received a $5.0M investment from Schneider Electric’s SE Ventures. UnitX offers a software-defined imaging system and edge AI compute platform for visual inspection and defect detection. For semiconductors, it provides inline monitoring for packaging, cutting, alignment, and wire bonding. It also has applications in PCBA, consumer electronics, lithium-ion battery, and automotive inspection. Founded in 2018, it is based in Santa Clara, California, USA.

Wireless

Falcomm drew $4.0M in funding from Squadra Ventures. Falcomm develops power amplifiers for wireless communications. Its first product targets 5G mmWave and sub-6GHz applications and supports Gb/s modulated signals with high spectral efficiency. The startup claims its technology provides high energy efficiency while using half the silicon area of commercially available Doherty amplifiers. A spin out from Georgia Institute of Technology founded in 2021, it is based in Atlanta, Georgia, USA.

Photonics & optics

Astrape Networks raised €1.6M (~$1.7M) in pre-seed funding from Brabantse Ontwikkelings Maatschappij, Rabobank, Brabant Startup Fonds, TTT Smart Industries, Shift Invest, and PhotonDelta. Astrape Networks offers photonic ICs for the data center. Its technology uses multi wavelength channels and advanced control signaling to transmit data through fast circuit switching. The approach allows for data to be only converted into optical signals at the edges of the network, eliminating the need for the intermediate optical/electrical/optical conversions at the I/Os of switches. Founded in 2022, it is based in Eindhoven, the Netherlands.

Quantum computing

Nanofiber Quantum Technologies (NanoQT) raised $8.5M in convertible notes from Phoenix Venture Partners, JAFCO Group, Sparx Group, Keio Innovation Initiative, and Waseda University Ventures. NanoQT is developing quantum hardware using nanofiber-based cavity quantum electrodynamics (QED). The hybrid system, which relies on the way light and matter interact, uses atoms to perform quantum computation and photons to rewrite and transmit information. The startup says its technology is scalable, allowing up to 10,000 atoms to be accommodated in a cavity. Additionally, the nanofiber resonators can be coupled with optical fibers for integration into a quantum network. Funds will be used for R&D and hiring. Founded in 2022 based on research from Waseda University, it is based in Tokyo, Japan.

Delft Circuits raised €6.3M (~$6.8M) in Series A funding led by DTXL, joined by existing investors including High-Tech Gründerfonds. The startup makes cryogenic cabling for quantum computers using ultrathin, multi-layer flexible superconducting circuits for data transfer into the quantum system. Founded in 2016, it is based in Delft, the Netherlands.

Orange Quantum Systems received €1.5M (~$1.6M) in pre-seed financing led by QDNL Participations and Cottonwood Technology Fund. Orange QS provides diagnostic systems for high-throughput quantum chip testing with the aim of reducing the test cost per qubit. Its system comprises software diagnostics libraries integrated with measurement and control hardware components. The platform will be further developed for superconducting devices and expanded to other qubit platforms. Funds will be used to develop its next generation of quantum chip test equipment. Founded in 2020, it is based in Delft, the Netherlands.

Sensors

MantiSpectra drew €4.0M (~$4.3M) in seed funding led by Innovation Industries and PhotonVentures. MantiSpectra makes spectral sensors on a chip for near-infrared spectroscopy. The device integrates filters and a multi-pixel array containing sixteen detectors with high responsivity in the NIR region from 850 to 1700 nm. It targets material composition analysis for applications such as agrifood production, sorting plastic for recycling, process analytics, and consumer devices. A spin off from the Eindhoven University of Technology founded in 2020, it is based in Eindhoven, the Netherlands.

Displays & AR/VR

Aledia drew €120.0M (~$127.3M) in funding from existing shareholders CEA Investissement, Supernova Invest, Braemar Energy Ventures, and Bpifrance. Aledia develops and manufactures microLED chips based on a 3D architecture using gallium-nitride (GaN)-on-silicon microwires, which it claims can enable displays with better performance at competitive prices compared to current OLED technologies. The technology is currently implemented on 200mm silicon wafers and will evolve towards 300mm silicon wafers for high-volume production. It targets products such as smartphones, tablets, VR/AR systems, smartwatches, laptops, TVs, and large displays. Founded in 2011 as a spin out from CEA-Leti, it is based in Grenoble, France.

Automotive

Conigital raised £500.0M (~$626.0M) in Series A+ financing. Conigital is developing a full-stack driverless vehicle platform for any vehicle type that combines hybrid explainable AI and simulation with remote monitoring and teleoperation. It retrofits and custom builds driverless vehicles for industrial and commercial fleets for various use cases, such as airports, ports, logistic yards, industrial sites, and campuses. Funds will be used for hiring, partner ecosystem development, and M&A. Founded in 2015, it is based in Birmingham, UK.

Envisics added $50.0M to its Series C round with investment from M&G Investments, Hyundai Mobis, GM Ventures, Tarsadia Investments, Van Tuyl Companies, Jaguar Land Rover’s InMotion Ventures, and Stellantis Ventures. Envisics develops dynamic holographic technology for automotive augmented reality head-up displays (AR-HUD). Its technology enables images to be displayed at multiple distances simultaneously and adapt to ambient light changes. Envisics says it combines algorithms, fewer but more robust optics, and high magnification designs to reduce waste heat and energy by directing light where it’s needed. Founded in 2018 as a spin out from AR wearables company DAQRI, it is based in Milton Keynes, UK.

Fernride added $19.0M to its Series A round with investment from Deep Tech and Climate Fonds, Munich Re Ventures, and Bayern Kapital, bringing the round to $50.0M. Fernride is working to automate transport logistics by gradually increasing the autonomy of trucks and tractors through a combination of autonomous driving technology and teleoperation. It equips trucks and tractor units with sensors and cameras so that they can be remote controlled via mobile networks. The company targets yard operations such as container handling, shunting, and hub-to-hub transportation. Funds will be used to scale operations and boost technology development. Founded in 2019 as a spin out from Technical University of Munich, it is based in Munich, Germany.

Batteries

Verkor raised at least €850.0M (~$912.1M) in Series C funding led by Macquarie Asset Management, joined by Meridiam, Renault Group, EQT Ventures, EIT InnoEnergy, Sibanye-Stillwater, Bpifrance, Crédit Agricole Assurances, Fonds Stratégique de Participations, CMA CGM Energy Fund, and Airbridge Investments. The funding comes alongside €650.0M (~$697.5M) in subsidies from France and €600.0M (~$643.8M) in debt support, announced in July, from the European Investment Bank. Verkor is a lithium-ion battery manufacturer that aims to reduce carbon impact by making electric vehicle batteries closer to where they’re used, implementing Industry 4.0 technologies, and automatically recycling production scrap. The funds will be used to establish a 16 GWh/year EV battery cell manufacturing plant in Dunkirk, France. Founded in 2020, it is based in Grenoble, France.

Ascend Elements drew $542.0M in Series D and other equity investment led by Decarbonization Partners, Temasek, and Qatar Investment Authority, joined by Tenaska, Alliance Resource Partners, CMA CGM Energy Fund, BHP Ventures, Fifth Wall, Hitachi Ventures, Mirae Asset, At One Ventures, Agave Partners, Alumni Ventures, and others. Ascend Elements produces lithium-ion battery cathode precursor and cathode active materials made from recycled battery feedstock using its Hydro-to-Cathode direct precursor synthesis process technology, which eliminates several intermediary steps in the traditional cathode manufacturing process. Funds will be used for construction of a manufacturing facility in Hopkinsville, Kentucky. Founded in 2015, it is based in Westborough, Massachusetts, USA.

Lyten raised $200.0M in Series B funding led by Prime Movers Lab, joined by Stellantis, FedEx Corporation, Honeywell, Walbridge Aldinger Company, and others. Lyten develops 3D graphene materials. The startup says the 3D graphene is more chemically and electrically reactive than 2D graphene while also being highly tunable for specific properties such as strength, hardness, and electrical conductivity due to its three-dimensional morphology. The production process, which uses methane as a feedstock, is carbon neutral at scale. It uses the material in its lithium-sulfur batteries, which the company says has over 2X the energy density of lithium-ion while being lightweight, resistant to thermal runaway, and made without nickel, cobalt, or manganese. Additionally, Lyten uses the 3D graphene material in sensor arrays with high sensitivity and selectivity for use in automotive, industrial, health, and safety applications, as well as in lightweight graphene-infused thermoplastics. It plans to begin shipping lithium-sulfur batteries in early 2024. Founded in 2015, it is based in San Jose, California, USA.

Table