Startup Funding: Q4 2024

The fourth quarter of 2024 saw five mega-rounds of over $100 million. One of the hottest areas continues to be AI hardware, with one company developing RISC-V AI processor IP bringing in nearly $700 million in the largest round for a chip design startup this year. It was far from the only AI startup, however, as several companies gathered support for neuromorphic designs focused on ultra-low power applications and others touted their generative AI prowess.

Interconnect technologies, key to both the build out of AI data centers and chiplet technologies, was another area that saw numerous large rounds, with investors betting that optical approaches can overcome bandwidth bottlenecks both inside the package and at rack scale.

Plus, AI in EDA, test chips, and a massive round for a packaging firm feature in this report that highlights 75 companies that collectively raised over $3 billion in Q4 2024.

• Summary table
• Chips
• AI hardware
• EDA
• Packaging & test
• Materials
• Memory & storage
• AMS, wireless & power
• Photonics & optics
• Quantum computing
• Previous reports

Chips

Enfabrica raised $115.0M in Series C funding led by Spark Capital and joined by Arm, Cisco Investments, Maverick Silicon, Samsung Catalyst Fund, VentureTech Alliance, Atreides Management, Alumni Ventures, IAG Capital, Liberty Global Ventures, Sutter Hill Ventures, and Valor Equity Partners. Enfabrica designs a high-bandwidth network interface controller fabric optimized for accelerated computing clusters. Its Accelerated Compute Fabric (ACF) NIC chips logically interconnect GPUs and accelerators across a high-performance scale-out network in an AI data center and provide multi-port 800-Gigabit-Ethernet connectivity to GPU servers. Funds will be used to drive the volume production ramp of its 3.2 Tbps ACF SuperNIC chip, which will be available in Q1 2025. Founded in 2019, it is based in Mountain View, California, USA.

Panmnesia drew over $60.0M in Series A funding led by InterVest and joined by Korea Investment Partners, KB Investment, Woori Venture Partners, BSK Investment, Nvestor, Murex Partners, Daesung Private Equity, TS Investment, Daekyo Investment, GNTech Venture Capital, SL Investment, TimeWorks Investment, Quantum Ventures Korea, and Yuanta Investment Korea. Panmnesia develops Compute Express Link (CXL) IP and switch chips. The startup recently introduced CXL 3.1 IP with two-digit nanosecond latency, a CXL-GPU storage expansion solution that can supply terabytes of memory to the GPU, and a CXL-enabled AI cluster for acceleration of retrieval-augmented generation. Funds will be used to manufacture its CXL 3.1 switch silicon chip and further develop its CXL IP. A spin out from KAIST founded in 2022, it is based in Seongnam, South Korea, and has raised over $80M to date.

Axiado raised $60.0M in Series C funding led by Maverick Silicon with participation from Samsung Catalyst Fund, Atreides Management, and Crosslink Capital. Axiado provides a trusted control/compute unit (TCU) that integrates key infrastructure security and management functions into an SoC. Its platform includes a root-of-trust/cryptography core and AI-driven pre-emptive threat detection engine to detect, quarantine, and protect systems against breaches, vulnerabilities, and attacks as they occur directly in hardware. It also includes a dynamic thermal management solution to predict and adjust the cooling requirements of AI servers. The funds will support go-to-market initiatives. Founded in 2017, it is based in San Jose, California, USA.

Azimuth AI received $11.5M in funding led by Cyient and Moneta Ventures, joined by Growth Factory and AUM Ventures. Azimuth AI produces highly customized SoCs and ASICs for smart city, 2/3-wheeler vehicles, and other edge and embedded computing markets. Funds will be used to productize its first-generation SoC for an OEM and expand its team. Founded in 2022, it is based in Folsom, California, USA.

Mindgrove Technologies raised $8.0M in Series A funding led by Rocketship.vc and Speciale Invest, with participation from Mela Ventures, Peak XV Partners, Whiteboard Capital, and individual investors. Mindgrove offers a high-performance security-focused microcontroller SoC for IoT devices such as smart watches, meters, locks, and access control units. The company is also developing a vision SoC for edge computing and vision processing applications in CCTV cameras, dashcams, video recorders, ADAS, and smart TVs. Its chips are built using the open-source Shakti processor cores developed at IIT Madras, which are based on the RISC-V ISA. Funds will be used for hiring and to accelerate production and sales of its first chip. Founded in 2021 and incubated at IIT Madras, it is based in Chennai, India.

Ubitium drew $3.7M in seed funding led by Runa Capital, Inflection.xyz, and KBC Focus Fund. Ubitium has developed a universal RISC-V processor that handles CPU, GPU, DSP, and FPGA computing workloads on a single chip. Its workload-agnostic microarchitecture allows the same transistors to be reused for different processing tasks ranging from simple control logic to massive parallel data flow processing, including AI compute. It initially targets embedded systems and robotics. Funds will be used to develop the first prototypes and prepare initial development kits, with the first chips planned for 2026. Founded in 2024, it is based in Düsseldorf, Germany.

AI hardware

Tenstorrent raised over $693.0M in Series D funding led by Samsung Securities and AFW Partners, joined by XTX Markets, Corner Capital, Protagonist Management, MESH Ventures, Export Development Canada, Healthcare of Ontario Pension Plan, LG Technology Ventures, Hyundai Motor Group, Fidelity Management & Research Company, Innovation Engine, Baillie Gifford, Bezos Expeditions, and others. Tenstorrent develops AI processor IP based on RISC-V. The processor supports a range of precision formats and is made up of a grid of cores that each include an array math unit for tensor operations, a SIMD unit for vector operations, a network-on-chip (NoC) to move data from core-to-core and chip-to-chip, five “baby RISC-V” processors to help direct the NoC, and up to 1.5MB of SRAM. Optimized for high performance per watt, it targets applications from server and HPC to edge cloud to edge devices. The company also provides an out-of-order, superscalar, 8-wide decode processor based on the RISC-V RVA23 profile that can scale down to a 2-wide implementation. Along with offering its IP, Tenstorrent also deploys it in chips, PCIe cards, and workstation and server systems. Funds will be used to build out open-source AI software stacks, hire developers, expand its global development and design centers, and build systems and clouds for AI developers. Founded in 2016.

HyperAccel raised KRW 55.0B (~$40.0M) in Series A funding led by Korea Investment Partners, joined by Korea Development Bank, KB Investment, Mirae Asset Venture Investment, SBVA, KDB Capital, BonAngels Venture Partners, Company K Partners, LB Investment, and Vickers Venture Partners. HyperAccel is developing processor IP and chips for generative AI inference. The startup says its LLM processing unit (LPU) has a latency-optimized, scalable architecture that balances memory bandwidth and compute logic to maintain effective bandwidth usage of ~90%. It employs a peer-to-peer communication technology to perform low-latency synchronization of multiple LPUs, enabling acceleration of LLMs with ten to one hundred billion parameters. Funds will be used to mass produce a 4nm LPU chip with LPDDR5X memory. Founded in 2023, it is based in Seoul, South Korea.

EdgeCortix received a JPY 4.0B (~$25.9M) subsidy from Japan’s New Energy and Industrial Technology Development Organization. EdgeCortix provides neural network accelerator IP based on a runtime reconfigurable architecture that reconfigures data paths between accelerator engines to achieve better parallelism and reduce on-chip memory bandwidth. Targeting edge inference applications, it supports both convolutional and transformer networks. The company has implemented the IP in an AI accelerator chip designed for applications requiring real-time Batch=1 AI inferencing. Also available as M.2 modules and PCIe cards, the hardware solutions are paired with a compiler and software framework for deploying edge AI applications. Funds will support development of a chiplet-based platform capable of providing both energy efficient AI inference and Radio Access Network (RAN) acceleration. Founded in 2019, it is based in Tokyo, Japan.

Gemesys raised €8.6M (~$9.1M) in pre-seed funding led by Amadeus APEX Technology Fund and Atlantic Labs, joined by NRW.BANK, Sony Innovation Fund, and Plug and Play Tech Center. Gemesys makes AI chips using memristors for fully analog neural network processing at the edge. The startup says its brain-inspired chip architecture can perform both on-device training and inference and can train neural networks with a fraction of the data required by conventional architectures. It also supports distributed capabilities such as outlier detection, data drift management, and model personalization. Funds will be used for R&D and hiring. A spin off from Ruhr University Bochum founded in 2021, it is based in Bochum, Germany.

Neuronova drew €1.5M (~$1.6M) pre-seed funding led by 360 Capital and CDP Venture Capital. Neuronova is developing ultra-low power processors that implement neurons and synapses as physical structures on silicon to emulate the brain directly at the hardware level for AI on battery-powered devices. Target markets range from consumer electronics to IoT. Funds will be used for hiring, R&D, and design and production of its first prototype processor. Founded in 2024, it is based in Milan, Italy.

Netrasemi drew INR 100.0M (~$1.2M) in seed financing led by Unicorn India Ventures. Netrasemi makes SoCs for real-time AI on edge IoT devices based on a power-efficient graph stream parallel architecture. The SoCs contain acceleration cores, a heterogeneous compute fabric, and a UCIe-based die-to-die interconnect. They are optimized for vision and smart sensor applications. Founded in 2020, it is based in Thiruvananthapuram, India.

EDA

Keysom raised €4.0M (~$4.4M) in funding from Bpifrance, IRDI Investment Capital, Xplore By Épopée Gestion, and EuroBIM. Keysom offers a ‘no-code’ platform for architectural exploration of custom RISC-V processors. The startup says its software allows exploration of up to 12 million processor combinations to produce a result that is PPA optimized for the specific application. Funds will be used for development, hiring, and commercial expansion. Founded in 2022, it is based in Pessac, France.

Alpha Design AI drew $3.1M in pre-seed funding led by ScOp Venture Capital, joined by Impact Assets, Amino Capital, and others. Alpha Design AI develops AI agents for EDA workflows with the aim of boosting RTL design, debugging, and verification productivity. The startup says the AI agent enables designers to transform concepts into precise design specifications using simple language prompts, analyzes and generates RTL design specs and code, auto-completes Verilog, automates the creation of testbenches, and autonomously verifies and debugs design code through real-time learning from simulations. Founded in 2024, it is based in Goleta, California, USA.

Packaging & test

SJ Semiconductor closed $700.0M in financing from Wuxi Chanfa Science and Technology Innovation Fund, Jiangyin Binjiang Chengyuan Investment Group, Fortera Capital, Shanghai International Group, Lingang Group, Zhongguancun Independent Innovation Fund, China Life Private Equity Investment Limited, and others. SJSemi is a 12-inch middle-end-of-line (MEOL) packaging foundry that specializes in advanced wafer bumping, 3D multi-die integration technology, wafer-level chip-scale packaging (WLCSP), wafer-level antenna-in-package, and testing. It offers a system-in-package platform for 3D heterogenous integration of active and passive devices that comprises Cu vertical interconnect, multi-layer double sided fine pitch RDLs, wafer-level multi-layer precise aligned antennas, flip-chip, and SMT. SJSemi also recently introduced a 3x reticle TSV interposer technology. Funds will be used to further develop its multi-die integration solutions. Founded in 2014, it is based in Jiangyin, China.

Chipmetrics was awarded a €1.0M (~$1.1M) grant from Business Finland. Chipmetrics makes metrology chips for semiconductor process control. It offers 3D ultra-high aspect ratio silicon test chips, along with test elements, test structures, and monitor wafers, for conformality measurements in material deposition processes such as atomic layer deposition and chemical vapor deposition. The chip can be used to compare different 3D thin film processes for both process development and production monitoring. Funds will support business development activities and international expansion. Founded in 2019 as a spin out from VTT Technical Research Centre, it is based in Joensuu, Finland.

Materials

Boston Materials received $13.5M in funding led by AccelR8 and Mitsubishi Chemical Group’s Diamond Edge Ventures with participation from Valo Ventures, Gatemore Venture Partners, Collab Fund, and Woori Venture Partners. Boston Materials uses vertically aligned carbon fibers to create materials with enhanced energy transfer properties for the semiconductor, aerospace, and EV industries. For semiconductors, it produces thermal interface materials as both liquid metal and adhesive films, the former of which has <3 mm2-K/W thermal resistance at 150µm bond line thickness and accommodates die warpage. Funds will be used to accelerate its expansion into the semiconductor market. Founded in 2016, it is based in Billerica, Massachusetts, USA.

Agnit Semiconductors drew $3.5M in seed funding led by 3one4 Capital and Zephyr Peacock. Agnit Semiconductors designs and manufactures gallium nitride (GaN) wafers and devices, primarily for RF applications. It focuses on the defense and telecommunications markets with plans to expand to consumer electronics and electric two-wheelers. Founded in 2019, it is based in Bengaluru, India.

Memory & storage

Biomemory received $18.0M in Series A financing led by Crédit Mutuel Innovation, with participation from Bpifrance, Paris Business Angels, Sorbonne Venture, Adnexus, Prunay, Next Sequence, and Accelerem. Biomemory is developing DNA data storage using a synthesis and copy process based on synthetic biology. The process produces long DNA fragments that can be stored as inert polymers for thousands of years without any energy input at data densities much higher than SSDs or magnetic tape. The company has produced a DNA storage card and plans to scale up to exabyte storage for data center usage. Funds will be used to complete the development of the first generation of their data storage appliance and hire. Founded in 2021, it is based in Paris, France.

AMS, wireless & power

Iontra drew $45.0M in Series C funding led by Volta Energy Technologies. Iontra makes RISC-V battery charge control MCUs for a range of consumer and industrial products, including power tools, smartphones, and wearables. The battery sensing and charging method protects the battery by minimizing plating and dendrite formation, which cause safety issues and limit battery performance. The company says its chemistry-agnostic technology can concurrently increase battery cycle life and charge speed up to and greater than 200% and support cold weather charging to -20°C without changing the design and chemistry of existing battery cells. Founded in 2013, it is based in Centennial, Colorado, USA, and has raised $120M to date.

Ookuma Diamond Device drew JPY 4.0B (~$26.8M) in pre-Series A financing led by Globis Capital Partners, joined by Coral Capital, aSTART, Green Co-Invest Investment, Japan Post Bank Spiral Regional Innovation Fund, Mitsui Sumitomo Insurance Venture Capital, SMBC Venture Capital, and others. Ookuma Diamond Device is developing diamond semiconductor amplifiers and power electronics. It is specifically developing devices to be used in the Fukushima Daiichi Nuclear Power Plant decommissioning project, as diamond-based devices can withstand the high temperatures and radiation of that environment. The startup is also exploring applications in 6G wireless communications base stations, satellite communication equipment, and radar. Funds will be used to construct a diamond semiconductor manufacturing facility, with operation slated to begin in 2026. A spin out from Hokkaido University founded in 2022, it is based in Sapporo, Japan.

Forefront RF received £16.0M (~$20.7M) in Series A funding from BGF, Foresight Group, Octopus Ventures, and Cambridge Innovation Capital. Forefront RF offers an RF front end module that incorporates a power amplifier, low noise amplifier, coupler, and a tunable duplexer that dynamically self-configures according to available frequencies and services needed. Its tunable duplexer replaces fixed-frequency filters and duplexers to save PCB space and employs a self-interference cancellation system that dynamically inverts and neutralizes unwanted signals to address leakage and interference issues. Its first-generation product, launching in 2026, targets the cellular-enabled smartwatch market. A spin out from the University of Bristol founded in 2020, it is based in Cambridge, UK.

Photonics & optics

Lightmatter raised $400.0M in Series D funding led by T. Rowe Price Associates with participation from existing investors including Fidelity Management & Research Company and Google Ventures. Lightmatter provides a 3D-stacked photonic interconnect and packaging technology platform for data center AI and HPC workloads. The package comprises a customer’s ASIC chiplets along with a reconfigurable optical interposer with integrated controller, bi-directional pluggable high density fiber array, and substrate. The photonics engine supports single-chip to multi-die complexes up to wafer-scale and is capable of connecting thousands to millions of processors. The company says that delivering IO in 3D increases AI cluster bandwidth and performance, while reducing power consumption and freeing up XPU shoreline to support more memory. Funds will be used to ready the technology for mass deployment in partner data centers. Founded in 2017 and based in Mountain View, California, USA, it has raised $850M to date.

Ayar Labs raised $155.0M in financing led by Advent Global Opportunities and Light Street Capital, joined by AMD Ventures, Intel Capital, Nvidia, 3M Ventures, Autopilot Ventures, Applied Ventures, Axial Partners, Boardman Bay Capital Management, GlobalFoundries, IAG Capital Partners, Lockheed Martin Ventures, Playground Global, and VentureTech Alliance. Ayar Labs develops in-package optical interconnects. Optimized for AI training and inference, the standards-based solution pairs optical I/O chiplets with multi-wavelength light sources to provide terabits of bandwidth bidirectionally from an ASIC/FPGA/XPU package with reach from mm to km package-to-package. The chiplet’s modular multiport design can carry eight light channels, while the light source provides up to 16 wavelengths of light and powers up to 16 ports. The company is preparing for high volume manufacturing. Founded in 2015, it is based in San Jose, California, USA, and has raised $370M to date.

Xscape Photonics raised $44.0M in a Series A round led by IAG Capital Partners with investment from Altair, Cisco Investments, Fathom Fund, Kyra Ventures, LifeX Ventures, Nvidia, and Osage University Partners. Xscape Photonics offers a fully programmable, multi-wavelength photonics platform for AI data center fabrics. The platform can handle hundreds of colors on a single fiber, which the startup claims increases bandwidth out of a GPU package by 10X, reduces power consumption, and improves overall inference performance for AI workloads. Funds will be used to accelerate development and hire. Founded in 2022, it is based in New York, New York, USA, and has raised $57M to date.

Oriole Networks received $22.0M in Series A financing led by Plural UK Management, joined by UCL Technology Fund, XTX Ventures, Clean Growth Fund, and Dorilton Ventures. Oriole Networks uses optical networks to create networks of AI chips that can increase the speed of large language model (LLM) training with reduced latency and increased energy efficiency. A spin out from University College London founded in 2023, it is based in London, UK.

Akhetonics received €6.0M (~$6.3M) in seed funding led by Matterwave Ventures, with participation from 468 Capital, Bayern Kapital, Runa Capital, Rheingau Founders, and others. Akhetonics develops an all-optical cross-domain processor for general-purpose, ultra-low power, HPC, and AI workloads. Its approach combines optical digital computing with optical analog computing and optical quantum computing to create components including logic gates, complex mathematical operations, feed-forward for continuous variables, optical memory, and analog-digital converters that allow them to interact. The startup has also created in-house design automation tools for a photonic RTL-to-GDSII flow as well as an all-optical control flow. Founded in 2021, it is based in Berlin, Germany.

Opticore launched with $5.0M in seed funding from Sagax Capital, Jetha Global, and others. Opticore is developing optical processing units (OPUs) that can perform the same parallelized computational tasks as GPUs, particularly matrix multiplication, using photonic logic based on photoelectric multiplication in coherent detection. Fabricated using standard foundry processes, the OPU is designed to handle memory-intensive AI tasks. It converts memory data to optical beams for on-chip movements with waveguides and computations with optical devices, eliminating the need for memory to be physically close to processors. The startup claims its approach overcomes the memory bottleneck while increasing compute density and lowering energy consumption by up to 100x in its prototype device. Funds will be used to build a larger chip. Founded in 2024, it is based in Fremont, California, USA.

Quantum computing

Quantum Brilliance drew $20.0M in Series A funding from the National Reconstruction Fund Corporation, Breakthrough Victoria, Main Sequence, Intervalley Ventures, In-Q-Tel, Alium Capital Management, Investible, and Jelix Ventures. Quantum Brilliance builds quantum computers based on nitrogen vacancy centers in synthetic diamonds that can operate at room temperature in any environment with long coherence times and a small form factor. The technology does not require cryogenics, vacuum systems, or precision laser arrays, which the company says enables its systems to be deployed on site or at the edge. In addition to large-scale quantum systems, the startup sees its technology being used as quantum accelerators in applications like satellites and automotive. Quantum Brilliance also offers a software development kit that includes high performance emulators for development and testing of quantum applications. Funds will be used to establish a foundry for quantum diamond devices. Founded in 2019 as a spin-out from the Australian National University, it is based in Sydney, Australia.

Lightsynq emerged from stealth with $18.0M in Series A funding led by Cerberus Ventures, with participation from Murata Electronics North America, IAG Capital Partners, Safar Partners, QVT, Element Six, and In-Q-Tel. Lightsynq is developing optical quantum interconnects that enable hardware providers to link quantum processors without redesign of existing hardware. The technology uses quantum memories based on color centers in diamond photonic devices, an approach the startup says provides a path to foundry-scale production. Funds will be used to build the first prototypes. Founded in 2024, it is based in Brighton, Massachusetts, USA.

Accelerators

The Northeast Microelectronics Coalition Hub awarded a total of $1.0M in grants split between 13 microelectronics startups with applications spanning power electronics, AI hardware, quantum technology, and wearable computing.

Silicon Catalyst added six new companies to its incubator program, including startups developing solutions for chiplet integration, RISC-V IP for algorithmic processing, and PUF post-quantum cryptographic ICs.

Table