AI chips and data center communications see big funding; 75 startups raise $2 billion.
The first quarter of 2025 saw six companies raise at least $100 million in investment. Of those, three went to quantum hardware companies, with major investment into neutral atom, superconducting, and hybrid quantum control approaches.
AI chips and enabling technology were another big winner in the quarter, with companies developing optical communications tech for chips and data center infrastructure pulling in over $400 million. Outside the data center, six startups are using in-memory compute technologies to lower the power requirements of on-device AI processing.
Several new companies emerged from stealth to bring coherent DSPs, a novel memory structure, and MEMS-photonics integration to market. They join makers of RISC-V processors, atomic layer processing and metrology equipment, and high-voltage power semiconductors in the 75 companies featured for Q1 2025 that collectively raised over $2 billion.
Retym emerged from stealth with $75.0M in Series D funding led by Spark Capital, joined by Kleiner Perkins, Mayfield, and Fidelity Investments. Retym specializes in programmable coherent DSP solutions for cloud and AI infrastructure. Its approach integrates high-performance DACs and ADCs with an intelligent DSP processor. The startup’s first chip, currently undergoing testing and validation, is optimized for coherent optics in the 30 km to 40 km range. Funds will support scaling to production and continued product development. Founded in 2021, it is based in Cupertino, California, USA, and has raised $180M to date.
Baya Systems raised over $36.0M in a Series B round led by Maverick Silicon, joined by Synopsys, Matrix Partners, and Intel Capital. Baya Systems offers a software-defined, unified fabric for SoCs and chiplets that provides a common transport with support for multiple protocols and coherency needs within a unified design flow. The startup’s software enables continuous refinement of data-driven architecture and micro-architecture development from initial specification through post-silicon tuning, with built-in simulation and workload analysis. Its portfolio of system IP includes non-coherent, cache coherent, and non-blocking switch fabrics, all built on a common transport. Funds will support operational growth, market expansion, and development. Founded in 2023, it is based in Santa Clara, California, USA.
AheadComputing secured $21.5M in seed funding led by Eclipse Ventures with participation from Maverick Capital, Fundomo, EPIQ Capital Group, and individual investor Jim Keller. AheadComputing designs 64-bit RISC-V application processor IP. The startup says it has developed a microprocessor architecture for general-purpose computing that also enables AI workloads to run efficiently. Aiming to serve server, client, mobile, and edge applications, the company is focused on producing high per-core performance while addressing thermal density constraints and multiprocessor scalability. Funds will be used for development and commercialization. Founded in 2024, it is based in Beaverton, Oregon, USA.
EnCharge AI secured over $100.0M in Series B funding led by Tiger Global with participation from Maverick Silicon, Capital TEN, SIP Global Partners, Zero Infinity Partners, CTBC VC, Vanderbilt University, Morgan Creek Digital, RTX Ventures, Anzu Partners, Scout Ventures, AlleyCorp, ACVC, S5V, Samsung Ventures, HH-CTBC, VentureTech Alliance, In-Q-Tel, Constellation Technology Ventures, and others. EnCharge develops analog in-memory computing AI chips for on-device inference. It utilizes charge-domain computation based on intrinsically precise metal capacitors that can be implemented in standard CMOS, which the startup claims overcomes signal-to-noise ratio limitations and variability in current-based analog IMC technology, while providing broad support for AI model types and operators. The company says its publicly disclosed silicon provides 150 TOPS/W at 8-bit precision, with its upcoming accelerator anticipated to reach 375 TOPS/W. EnCharge plans to implement the technology in a range of form factors, including chiplets, ASICs, and standard form-factor PCIe cards, and will also provide software for orchestration between on-device and cloud deployments. Funds will support the commercialization of its first client computing-focused AI accelerator products in 2025 and progress along its product roadmap. Founded in 2022 with technology developed at Princeton University, it is based in Santa Clara, California, USA, and has raised $144M to date.
Axelera AI secured a grant of up to €61.6M (~$66.5M) from the EuroHPC Joint Undertaking and member states as part of the Digital Autonomy with RISC-V for Europe (DARE) Project. Axelera AI provides an AI acceleration platform for generative AI and computer vision inference that utilizes SRAM-based digital in-memory computing and a RISC-V controlled dataflow architecture to minimize data movement between memory and compute elements. It recently announced a scalable chiplet containing its technology that targets markets such as HPC, enterprise data centers, robotics, and automotive. The company also offers quantization techniques and mapping tools that reduce AI computational load and increase energy efficiency while minimizing accuracy loss for a wide range of networks. Founded in 2021, it is based in Eindhoven, the Netherlands.
MemryX raised $44.0M in Series B funding. MemryX makes accelerator chips for vision-based AI inference at the edge. It uses compute-at-memory technology and a configurable dataflow architecture. It uses memory as the only interconnect between compute engines, rather than managing data movement with a control plane or a network-on-chip. Up to 16 chips can work together as one logical unit with a single connection to the host processor. Applications include video management systems, industrial PCs, and edge servers. Funds will be used to ramp production of its current 6 TFLOPs accelerator chip and complete the design of its upcoming one. Founded in 2019, it is based in Ann Arbor, Michigan, USA.
Positron secured $23.5M in funding from Flume Ventures, Valor Equity Partners, Atreides Management, Resilience Reserve, and others. Positron designs energy-efficient hardware for large language model (LLM) inference. Its FPGA-powered servers are based on a memory-optimized architecture that achieves over 93% bandwidth utilization and can support trillion-parameter models. The company says it can map any transformer model directly onto hardware. Funds will be used to scale production of its chips. Founded in 2023, it is based in Reno, Nevada, USA.
Vertical Compute launched with €20.0M (~$20.5M) in seed funding led by imec.xpand, joined by Eurazeo, XAnge, Vector Gestion, and imec. Vertical Compute is developing a vertically integrated memory and compute technology. Targeted for AI hardware accelerators, its chiplet-based solutions store bits in a high-aspect-ratio vertical structure that integrates vertical data lanes on top of computation units. By reducing data movement and bringing large data closer to computation, the startup says its approach significantly reduces power compared to DRAM while improving privacy by eliminating the need for remote data transfers. A spin-off from imec founded in 2024, it is based in Louvain-La-Neuve, Belgium.
Axion Technologies received KRW 2.0B (~$1.5M) in seed funding led by LinkBricks Ventures and Betterground. Axion offers AI-driven chip design optimization software, including automatic generation of customized cell libraries and a design for manufacturing (DFM) tool. Founded in 2024, it is based in Seongnam, South Korea.
GS Microelectronics (GSME) secured $18.0M in growth investment. GSME provides customized silicon solutions, including RF design, power management ICs, mmWave front-end IP, turnkey manufacturing operations, and quality assurance to IC design and system companies. Funds will support the launch of the company’s manufacturing dashboard platform that provides real-time visibility and control over the manufacturing process, enabling proactive monitoring, parametric analysis, and fault detection. Founded in 2021, it is based in San Jose, California, USA.
Atlant 3D raised $15.0M in a Series A+ funding round led by West Hill Capital. Atlant 3D develops direct atomic layer processing (DALP) equipment that integrates temporal and spatial atomic layer processing into a single system capable of creating complex, multi-material structures for microelectronics, semiconductors, optics, and other advanced devices in a single step. The technology uses microchemical reactors to deposit metals, semiconductors, and complex oxides exactly where needed, an approach that the company claims reduces material waste by 90%. Atlant 3D currently offers a DALP system for materials research, device R&D, and prototype development, with an industrial multimodular version in the works. Founded in 2018, it is based in Copenhagen, Denmark.
Clas-SiC Wafer Fab received a £12.0M (~$14.9M) investment from Archean Chemical Industries. Clas-SiC is a silicon carbide (SiC) wafer foundry specializing in prototyping, low- to medium-rate production, accelerated R&D cycle times, and subcontract process services. It is capable of processing 150mm SiC wafers through JBS/MPS diode and MOSFET flows. Key markets for its power semiconductors include electric vehicles and the renewable energy sector. Funds will be spent on capital equipment. Founded in 2017, it is based in Loghgelly, UK.
Wooptix raised over €10.0M (~$10.5M) in Series C funding led by Samsung Venture Investment Corporation and Spanish Society for Technological Transformation with participation from existing investors including Bullnet Capital, the European Innovation Council, Intel Capital, Mondragon Corporation, and TEL Venture Capital. Wooptix develops semiconductor wafer metrology equipment incorporating wavefront phase imaging. The startup says its technology enables blank and patterned wafer-shape measurement, including shape uniformity, nanotopography, and roughness, in less time and with higher resolution. It currently offers a manually operated 200mm and 300mm wafer metrology tool capable of collecting over 16 million data points with sub-nanometer height resolution and near real-time process capability for in-line factory measurements. It plans to introduce a fully automated tool this year. Founded in 2016, it is based in Tenerife, Spain.
Atmosic Technologies raised $40.0M in Series D funding led by Sutter Hill Ventures and joined by Clear Ventures, Quantum Innovation Fund, and others. Atmosic develops ultra-low-power Bluetooth and 802.15.4 (LR-WPAN) SoCs that utilize energy harvesting to enhance efficiency and reduce the need for battery replacements in wireless connected devices. To reduce power, it uses a sensor hub hardware block that allows data to be managed while the majority of the SoC, including the MCU, is in a hibernated state, along with an on-demand wakeup feature. Several of its products feature on-chip RF energy harvesting and can support various energy harvesting technologies, including photovoltaic, thermal, and motion-based harvesting. Funds will be used to scale manufacturing capacity and accelerate the development of its next-generation compute and connectivity platform. Founded in 2016, it is based in Campbell, California, USA.
Emtar Technologies closed its CAD $18.5M (~$13.0M) angel round. Emtar Tech is an IC design company specializing in wireless SoCs that integrate high-speed I/O, wireless connectivity, and intelligent self-computing capabilities for edge computing applications. It is working on developing high-performance, energy-efficient, and ultra-reliable fast response signal acquisition solutions. Founded in 2023, it is based in Richmond, British Columbia, Canada.
PseudolithIC raised $6.0M in seed funding led by Entrada Ventures, joined by Foothill Ventures and Uncork Capital. PseudolithIC makes RF chipsets using a manufacturing process that integrates compound semiconductor chiplets with silicon wafers, an approach it says delivers the performance of a monolithic microwave IC (MMIC) at a lower cost. Its initial products are millimeter-wave power amplifiers, low-noise amplifiers, and transmit/receive front-ends for markets such as terrestrial and satellite communications, aerospace, and IoT. Funds will be used for development and commercialization. Founded in 2021, it is based in Santa Barbara, California, USA.
Oso Semiconductor drew $5.2M in a seed round led by Engine Ventures, joined by Entrada Ventures, Berkeley SkyDeck, and J-Ventures. Oso Semi designs ultra-low-loss beamforming chips for phased array antennas in satellite communications (SATCOM), 5G, and radar systems. The startup has developed algorithmic approaches to phase shifting and combining that allow for a simplified chip design with fewer amplifiers, which it says provides power and cost savings over existing electrically steerable antennas. Funds will be used to build pre-production chipsets and evaluation systems for initial customers. Founded in 2022 as a spin-out from the University of California Berkeley, it is based in Mountain View, California, USA.
Cambridge GaN Devices (CGD) closed a $32.0M Series C round led by a strategic investor with participation from British Patient Capital, Parkwalk Advisors, BGF, Cambridge Innovation Capital, Foresight Group, and IQ Capital. CGD makes gallium nitride (GaN) power devices for high-power industrial, data center, and automotive markets. The company’s enhancement-mode GaN transistor can be operated like a silicon MOSFET without the need for special gate drivers, driving circuitry, or unique gate voltage clamping mechanisms, enabling a monolithic chip in a low-profile/low-inductance package. CGD recently introduced a combo device that combines GaN HEMT ICs and IGBTs in the same module or IPM for EV powertrain applications over 100kW. Funds will be used to expand operations. A spin-out from Cambridge University founded in 2016, it is based in Cambridge, UK.
Pakal Technologies raised $25.0M in Series B funding led by New Science Ventures with participation from Translink Capital and Arborview Capital. Pakal has developed an IGTO(t) silicon power switch for high-voltage applications. A direct drop-in replacement for IGBT, the IGTO(t) incorporates a novel trench structure and differentiated operational physics, which the company claims provides efficiency gains compared to IGBT. The startup expects to begin mass production of 650 V devices in early 2025, with 1200 V and 1700 V product lines to follow. Applications include EVs, data centers, solar and wind energy, battery/energy storage, and induction heating. Founded in 2017, it is based in San Francisco, California, USA.
Celestial AI raised $250.0M in Series C1 funding led by Fidelity Management & Research Company with participation from BlackRock, Maverick Silicon, Tiger Global Management, AMD Ventures, Koch Disruptive Technologies, Temasek, Xora Innovation, Porsche Automobil Holding, The Engine Ventures, and individual investor Lip-Bu Tan. Celestial AI offers a photonic technology that enables optically addressable memory and compute both within package and package-to-package across multiple racks. Its optical interconnect is not restricted by the package beachfront and can deliver data to any location on the compute die. The company claims its photonic fabric provides compute-to-compute and compute-to-memory connectivity at terabits per second bandwidth with nanosecond latencies, enabling the disaggregation of compute and memory in hyperscale data centers. The technology is compatible with standards including CXL, PCIe, UCIe, HBM, and proprietary electrical communication links, as well as with industry-standard manufacturing and 2.5D packaging processes. Founded in 2020, it is based in Santa Clara, California, USA, and has raised $515M to date.
Lumotive closed a $45.0M Series B round with investment from Swisscom Ventures, East Bridge Capital, EDOM Technology, Grazia Equity, Hokuyo Capital, TSVC, Gates Frontier Fund, MetaVC Partners, Quan Funds, USAA, and Himax Technologies. Lumotive creates programmable optical semiconductor products using software-defined solid-state photonic beamforming technology. Its Light Control Metasurface chips, which are manufactured using conventional silicon fabrication processes, enable precise manipulation of light at a sub-micron level by dynamically shaping and steering light on the chip’s surface without relying on moving parts. With a native 160° steering field of view, applications range from 3D sensing and lidar to high-speed communications and optical switching. Funds will be used to expand to new markets, particularly AI data center infrastructure and satellite communications. Founded in 2017, it is based in Redmond, Washington, USA.
Salience Labs closed a $30.0M Series A round led by ICM HPQC Fund and Applied Ventures, joined by Braavos, Oxford Sciences Enterprises, Cambridge Innovation Capital, Silicon Catalyst, individual investor Jalal Bagherli, and others. Salience Labs develops photonic switch chips that enable high-speed, ultra-low latency networking fabrics for AI data center infrastructure. The photonic switch provides multi-hop large cluster connectivity and is compatible with existing infrastructure. Founded in 2021, it is based in Oxford, UK.
CamGraPhIC completed a €25.0M (~$27.1M) Series A round led by CDP Venture Capital, NATO Innovation Fund, Sony Innovation Fund, and Join Capital, with participation from Bosch Ventures, Frontier IP Group, and Indaco Venture Partners. CamGraPhIC develops graphene-based transceivers for data centers and telecom. The company claims its graphene integrated photonics technology provides higher bandwidth density and better latency performance while consuming 80% less energy compared to traditional pluggable data center optical transceivers. The production process is compatible with commercial semiconductor and photonics foundries. It initially targets AI GPU to HBM interconnects, with plans to expand to avionics, ADAS, and space applications. Funds will be used for R&D and to establish a pilot manufacturing line. Founded in 2018 as a spinoff from the University of Cambridge, it is based in Cambridge, England, UK.
iPronics raised €20.0M (~$20.8M) in Series A funding led by Triatomic Capital with participation from Fine Structure Ventures, Bosch Ventures, Amadeus Capital Partners, and Criteria Venture Tech. iPronics offers an optical networking engine targeted at AI data centers. Its optically switched fabric enables on-the-fly topology adaptation and extends programmability to physical layer connections. The technology uses a network of programmable unit cells in a waveguide mesh arrangement that route and process the light signal. Each programmable unit cell is capable of acting as a tunable coupler, combiner, variable optical attenuator, waveguide, or phase shifter. Funds will support market deployment of its optical circuit switch. A spinoff of the Technical University of Valencia founded in 2019, it is based in Valencia, Spain.
Hyperlume drew $12.5M in seed funding led by BDC Capital and ArcTern Ventures with participation from MUUS Climate Partners, SOSV, Intel Capital, and LG Technology Ventures. Hyperlume makes microLED-based optical interconnects for chip-to-chip and rack-to-rack communications in AI data centers and high-performance computing systems. The startup claims its interconnects, compatible with CMOS packaging and available as pluggable, mid-board, and co-packaged optics, provide high bandwidth, low latency, and reduced power consumption. Funds will be used to accelerate development, hire, and prepare for production to meet the demand for 800G and 1.6T interconnects. Founded in 2022, it is based in Ottawa, Ontario, Canada.
InfiniLink closed a $10.0M seed round from MediaTek, Sukna Ventures, Egypt Ventures, and M Empire Angels. InfiniLink develops silicon photonics integrated optical transceiver chiplets and SerDes that enable low-power pluggable transceiver modules and high-bandwidth-density co-packaged optical engines for AI data centers. Funds will be used to accelerate commercialization. Founded in 2022, it is based in Cairo, Egypt.
Photon IP drew €4.8M (~$4.8M) in seed funding led by Innovation Industries with participation from Faber, Brabantse Ontwikkelings Maatschappij, and PhotonDelta. Photon IP has developed advanced photonic integration technology to combine silicon and multiple optimized III-V materials, including indium phosphide for lasers, high-speed modulators, and photodetectors, onto a single photonic chip. The startup says its solution simplifies the manufacturing process for optical chips and opens a path towards mass deployment of advanced low-power optical engines for a wide range of applications. Founded in 2020, it is based in Eindhoven, the Netherlands.
QuEra Computing secured $230.0M in financing from Google, SoftBank Vision Fund, Valor Equity Partners, QVT Family Office, Safar Partners, Sabanci Holding, and others. QuEra is developing large-scale, fault-tolerant quantum computers based on neutral-atom qubits. Its technology is built on large-scale arrays of Rubidium atoms, laser-cooled to microkelvin temperatures and held in a vacuum by movable laser tweezers. These atoms serve as the memory banks for quantum computing with ultrafast, high-fidelity gates. The company offers cloud access to its first-generation analog 256-qubit processor and has a technology roadmap to millions of qubits. Funds will be used to accelerate development, production, and hiring. Founded in 2018 based on research conducted at Harvard University and MIT, it is based in Boston, Massachusetts, USA.
Quantum Machines raised $170.0M in Series C funding led by PSG Equity with participation from Intel Capital, Red Dot Capital Partners, and existing investors. Quantum Machines provides hybrid quantum control solutions that integrate quantum and classical computing operations, enabling precise control, real-time data processing, and complex algorithm orchestration. Its control systems are based on hybrid processing units, 16-core classical processors that can simultaneously generate and manipulate pulses to control, receive, and process data from qubits and execute real-time Turing-complete classical calculations. The platform supports any type of quantum processor. Founded in 2018, it is based in Tel Aviv, Israel, and has raised $280M to date.
Alice & Bob closed a €100.0M (~$104.2M) Series B round led by Future French Champions, AXA Venture Partners, and Bpifrance, joined by Elaia Partners, Breega, Supernova Invest, and the European Innovation Council. Alice & Bob is working to build fault-tolerant quantum computers based on superconducting cat qubits that inherently suppress bit-flip errors. By protecting against bit-flips, error correction only needs to correct phase-flips, which the company says reduces the number of physical qubits needed to make a logical qubit. It hopes to build a useful quantum computer by 2030. Funds will be used to finance the construction of a lab and production facility to enhance the performance of its system, improve error correction, and create its first error-corrected logical qubit. Based in Paris, France, it was founded in 2020.
Omnitron Sensors secured over $13.0M in Series A funding led by Corriente Advisors with participation from L’Attitude Ventures. Omnitron Sensors develops MEMS fabrication IP. Its approach uses silicon on insulator (SOI) and surface machining to create a hybrid design that can have three dimensions of polysilicon for the MEMS sensor and the actuator. Omnitron’s first product is a MEMS step-scanning mirror. Applications include optical cross-connects for AI data centers, optical subsystems in long-range lidar for autonomous navigation, see-through displays in extended reality headsets and eyewear, and laser spectrometry for methane gas detection. Founded in 2019, it is based in Los Angeles, California, USA.
Phlux Technology raised £9.0M (~$11.7M) in Series A funding led by BGF, joined by Octopus Ventures, Northern Gritstone, and Foresight Group. Phlux develops antimonide-based semiconductor infrared sensors for optical communications, lidar, and laser range finders. The startup claims its ultra-low-noise InGaAs avalanche photodiodes can improve data rates for free-space and fiber-optic communication systems with sensitivity improvements of 5dB or more compared to current commercial devices. Other applications include ADAS, industrial automation, defense and security, and high-precision gas sensing. Funds will be used for hiring, ramping up production, and accelerating the commercial launch of new product lines. A spin-out of Sheffield University founded in 2020, it is based in Sheffield, UK.
Zero Point Motion emerged from stealth with £4.0M (~$5.2M) in pre-Series A funding from SCVC, Foresight Group, Verve Ventures, and u-blox. Zero Point Motion develops positioning and navigation sensors that combine cavity optomechanics, silicon photonics, and MEMS. The startup’s initial focus is inertial sensors, namely accelerometers and gyroscopes, that use a photonic signal instead of the typical noisy electrical readout to improve sensitivity and reliability. Target applications include defense, space exploration, and autonomous off-highway vehicles, drones, agritech, and industrial systems. Founded in 2020, it is based in Bristol, UK.
Arm and Silicon Catalyst announced this year’s Arm Silicon Startups Contest. Open to early-stage startups building an SoC utilizing Arm processor technologies, the contest awards the winners with Arm technology credits and access to resources. Applications are open through April 25, 2025.
The SEMI opened its Startups for Sustainable Semiconductors Program, which gives startups developing sustainability-related solutions mentorship and a chance to pitch their ideas at SEMICON West. Focus areas include data center efficiency, alternative materials discovery, and energy-efficient chip design. Applications are open until May 1, 2025.
Siemens launched a startup program that will provide early-stage engineering and manufacturing startups with reduced-cost access to Siemens software and hardware.
Intel will spin out its venture capital arm, Intel Capital, into a standalone fund under a new name. Intel will remain an anchor investor.
The ChipStart UK incubator program announced its latest cohort of semiconductor startups.
Leave a Reply