本文由半导体产业纵横（ID：ICVIEWS）编译自semi engineering
Investors began to focus on interconnect technology, with 39 companies receiving $1.1 billion.
Data transmission is increasingly a focus for chip and system designers, and investors are starting to pay attention. A number of startups developing interconnect technologies that received strong support in March, including chiplet-enabled PHYs, photonic structures for dissolving computing and memory, and telecom transceiver modules.
In addition, there are new startups that are working to solve problems such as data stream processor architectures, custom deep learning chips, and optical networks connected to GPUs.
In addition, the large-scale funding round for ADAS/AD development tools, diamond semiconductors, more AI chips and nanomaterial integration included 39 startups, which collectively raised more than $1.1 billion in funding in March.
chip
Eliyan raised $60 million in a Series B funding round led by Samsung Catalyst Fund and Tiger Global Management, with existing investors including Intel Capital, SK Hynix, Cleveland Avenue and Mesh Ventures. Eliyan has developed chip-to-chip interconnect technology for multi-chip architectures in advanced packaging or standard organic substrates. Its NuLink technology is a superset of the Bunch of Wires (BoW) and Universal Chiplet Interconnect Express (UCIe), a chip-to-chip PHY for connecting different functions in the same package. It is capable of simultaneous bidirectional signal transmission, enabling the design to receive and transmit data simultaneously on the same line. The PHY was recently taped out on TSMC's 3nm process and uses a standard package with a bandwidth of 64G bps/bump. In addition to chip-to-chip, it provides chip-to-memory interconnects with bidirectional transceivers for each data channel. The common memory interface proposed by Eliyan will enable a single ASIC SKU to be configured to connect to different memory chiplets and non-memory chips, such as co-packaged optics and SerDes chiplets. The company is headquartered in Santa Clara, California, USA, and was founded in 2021.
Efficient Computer was selected with a $16 million seed round led by EclipseVC. Efficient Computer has developed a reconfigurable stream processor architecture and software stack that is claimed to be up to 100 times more energy-efficient than general-purpose CPUs on the market today. The Fabric architecture expresses a program as a "circuit" of instructions, showing which instructions communicate with each other, enabling the circuit to be spatially arranged on an extremely simple array of processors to execute the program in parallel. The startup has implemented the architecture in a test SoC. General-purpose processors can be programmed through mainstream embedded programming languages and AI/ML frameworks for a range of edge applications, including IoT, wearable and implantable health devices, space systems, and security and defense. The company was founded in 2022 and is headquartered in Pittsburgh, Pennsylvania, United States.
Artificial intelligence hardware
Taalas stood out with a $50 million funding round in two rounds led by Quiet Capital and individual investor Pierre Lamond. Taalas is developing an automated process for the rapid implementation of all types of deep learning models, including transformers, SSMs, diffusers, and MoE, in chips. The startup says the technology enables one of its chips to hold an entire large AI model without the need for external memory, dramatically increasing efficiency and increasing model size by a factor of 10-100. The company will tape out its first large language model (LLM) chip in the third quarter of 2024. The company was founded in 2023 and is headquartered in Toronto, Ontario, Canada.
NeuReality raised $20 million from the European Innovation Council Fund, Varana Capital, Cleveland Avenue, XT Hi-Tech, and Our Crowd, with participation from Cardumen Capital, Glory Ventures, and Alumni Venture Group. NeuReality builds purpose-built AI platforms optimized for deep learning inference in applications such as computer vision, natural language processing, recommendation engines, and generative AI in data centers and near-edge local locations. The company says its Network Addressable Processing Unit (NAPU) reduces reliance on CPUs, NICs and PCI switches by moving simple but critical datapath functions from software to hardware. Its inference PCIe card connects directly to Ethernet to manage AI queries from a massive data pipeline of millions of users and billions of devices. The company also provides software and runtime libraries to accelerate AI deployments. Funding will support the deployment of its inference solution. Founded in 2019 and headquartered in Caesarea, Israel, the company has raised $70 million to date.
Innatera Nanosystems raised €15 million ($16.3 million) in a Series A funding round from the Invest-NL Deep Tech Fund, the European Innovation Council Fund, MIG Capital, Matterwave Ventures and Delft Enterprises. Innatera has developed neuromorphic processors based on an analog mixed-signal computing architecture that mimics the brain's mechanisms for processing sensory data. Innatera's processors feature spiking neural networks with ultra-low power consumption and short response delays, with always-on pattern recognition in applications at the edge of the sensor. The funds will be used to accelerate mass production and expand the scope of applications. The company was founded in 2018 as a branch of Delft University of Technology and is headquartered in Reisvijk, the Netherlands.
Manufacturing, equipment and materials
Elephantech received 3 billion yen (about $19.8 million) in the Series E financing. Elephantech offers a printed circuit board manufacturing technology based on metal inkjet printing, which is said to be environmentally friendly, reducing CO2 emissions by 77% and water consumption by 95% compared to traditional manufacturing methods. The company says its technology can be applied to almost all PCB types by printing metal onto substrates only where necessary. At present, it mainly produces single-sided flexible substrates, and plans to produce multilayer and rigid PCBs in the future. It also plans to enhance its ability to print and use recyclable materials on biomass substrates. The funds will be used for commercialization. It was founded in 2014 and is headquartered in Tokyo, Japan, and has raised about 15B yen to date.
Diamfab raised €8.7 million (about $9.4 million) in the first round of funding from Asterion Ventures, Bpifrance, Kreaxi, Better Angle, Hello Tomorrow and Grenoble Alpes Métropole. Diamfab uses its epitaxial growth and doping synthetic diamond technology to synthesize value-added semiconductor and power stack wafers. It has also designed diamond electronic components that claim to be lighter and more energy-efficient than silicon devices, with characteristics that are resistant to high temperatures, radiation, and voltage. It targets the automotive, renewable energy and quantum industries, and the funds will be used to build a pilot production line and accelerate development. Diamfab was founded in 2019 as a branch of the CNRS laboratory Institut Néel and is headquartered in Grenoble, France.
Chiral Nano raised $3.8 million in seed funding led by Founderful and HCVC, as well as grants from ETH Zurich and Venture Kick. Chiral Nano is developing transistors based on nanomaterials, including transcarbon nanotubes. It designs a scalable manufacturing process to grow, select, and automatically integrate nanomaterials, such as carbon nanotubes or graphene, into transistors with rapid, high-quality, and full process control. Choral said its nanotransistor technology is suitable for quantum processors, low-power sensors and high-performance transistors. In addition to robotic assembly equipment for placing micro- and nano-sized materials, the startup also offers casting services. The funding will support the development of its next-generation robots. It was founded in 2023 as a spin-off from ETH Zurich and Empa, and is headquartered in Zurich, Switzerland.
Gaianixx added another 350 million yen ($2.4 million) to its previous 1 billion yen Series B funding round with investors including JIC Venture Growth Investments, Alconix Corporation and SMBC Venture Capital, bringing the round to 1.35 billion yen. Gaianixx has developed an epitaxial technique that utilizes a dynamic lattice matching mechanism to grow high-quality single crystals in multilayer structures. Its versatile interlayer technology can be applied to semiconductor wafers, including MEMS, power devices, and LEDs, as well as piezoelectric and ceramic films. The funds will be used for R&D, mass production and recruitment. It was founded in 2021 as a spin-off from the University of Tokyo and is headquartered in Tokyo, Japan.
Membrion will receive up to $100,000 in pilot project funding from 100+ Accelerator. The company, which manufactures electroceramic desalination membranes for semiconductor and industrial wastewater treatment, claims that its membranes can recover up to 98% of water under harsh conditions, such as high temperatures, low pH ranges, and challenging trace chemicals. The company was founded in 2016 and is headquartered in Seattle, Washington, USA.
Test, Measurement & Inspection
Robovision raised $42 million in a Series A funding round led by Target Global and Astanor Ventures, with participation from Red River West. Robovision provides deep learning computer vision platforms for a range of industries. For semiconductors, it can interface with DUVs, electron beams, and optical solutions to perform in-line automated defect classification (ADC) on wafers at sub-10nm nodes. The company was founded in 2012 and is headquartered in Ghent, Belgium.
Power devices
e-peas Semiconductors raised €17.5 million ($19 million) in funding led by Otium Capital, with participation from new investors Nomainvest and EICFund, as well as existing investors KBC Focus Fund, The Faktory, Wallonie Entreprendre, Noshaq-Leansquare and Invest BW. e-PEAS provides ultra-low-power management ICs for energy harvesting that collect and manage energy from environmental sources such as light, RF, temperature gradients, or vibration. It also offers ultra-low-power microcontrollers and imagers. The company was founded in 2014 and is headquartered in Leuven-neuve, Belgium.
Photonics and Optics
Celestial AI attracted $175 million in Series C funding led by the U.S. Innovation and Technology Fund, which also included AMD Ventures, Koch Disruptive Technologies, Temasek and its subsidiary XoraInnovation, IAG Capital Partners, Samsung Catalyst, and Smart Global Holdings, Porsche Automobil Holding, Engine Ventures, M-Ventures and Tyche Partners. Celestial AI provides a photonic technology that enables optically addressed storage and computation within and between chips. Its optical interconnects are not limited by packaging and can transmit data to any location on the computing chip. The company claims that its photonic network provides compute and compute and compute and storage connectivity with terabits per second bandwidth, nanosecond latency, and enables the decoupling of compute and storage in hyperscale data centers. The technology is compatible with standards including CXL, PCIe, UCIe, HBM, and proprietary telecommunications links. Funds will be used to execute client partnership projects and commercialization. The company was founded in 2020 and is headquartered in Santa Clara, California, United States.
EFFECT Photonics raised $38 million in Series D funding, led by the Innovation Industries Strategy Partners Fund and Invest-NL Deep Tech Fund, with participation from other existing investors. EFFECT Photonics designs photonic optics and pluggable transceiver modules for high-speed optical fibers in telecommunications and datacom networks. The technology uses an integrated optical SoC with digital signal processing and forward error correction combined with an ultra-pure, tunable light source. The funds will be used to accelerate development, commercialization and production. It is a spin-off of the Technical University of Eindhoven, founded in 2010 and headquartered in Eindhoven, the Netherlands.
Oriole Networks has received £10 million ($12.6 million) in seed funding led by UCL Technology Fund, Clean Growth Fund, XTX Ventures and Dorilton Ventures. Oriole Networks uses an optical network to connect thousands of GPUs together to increase the speed of large language model (LLM) training while reducing latency and improving energy efficiency. It was spun off from University College London and was founded in 2023 and is headquartered in London, England.
Quintessent raised $11.5 million in seed funding, led by Osage University Partners, with participation from M Ventures and existing investors Sierra Ventures, Foothill Ventures and Entrada Ventures. Quintessent develops optical interconnects based on multi-wavelength quantum dot comb lasers as part of a heterogeneous silicon photonics technology stack. The startup claims that the interconnection of its data center computing and switching chips reduces power consumption and the number of components required, while exponentially increasing bandwidth density. The company was founded in 2019 and is headquartered in Santa Barbara, California, United States.
Quantum computing
Maybell Quantum raised $25 million in a Series A funding round led by Cerberus Ventures. Maybell manufactures cryogenic dilution chillers and wiring solutions for quantum computer infrastructure. Funds will be used to expand production and open new production facilities. It was founded in 2021 and is headquartered in Denver, Colorado, United States.
sensor
SteerLight has secured €3.2 million ($3.5 million) in equity and loan financing, led by Stellantis Ventures, as well as Quantonation, LIFTT and Bpifrance. SteerLight develops a monolithic silicon photonic LiDAR chip that integrates a laser and detection mechanism. Its initial focus was on industrial applications, including autonomous robotics, with plans to expand into infrastructure solutions, smart cities, and mobility. Founded in 2022, the company is a subsidiary of CEA-Leti and is headquartered in Grenoble, France.
Displays and AR/VR
MICLEDI Microdisplays raised $25 million in a Series A funding round for imec.xpand, PMV, IMEC, KBC and SFPIM. MICLEDI manufactures microLED arrays and display modules for the AR market, including smart glasses and automotive HUDs. Its product portfolio includes gallium nitride-based blue, green, and red microLED arrays and AlInGaP-based red microLED arrays. Funds will be used to design and build active backplane ASICs, create fully functional microLED display modules, and recruit. The company was founded in 2019 as a spin-off of imec and is headquartered in Lüven, Belgium.
Car
Applied Intuition raised $250 million in a Series E funding round led by Porsche Investments Management and individual investors Bilal Zuberi and Elad Gil, with participation from Andreessen Horowitz, General Catalyst and other retail investors. Applied Intuition provides a development platform for developing and testing ADAS and autonomous driving systems in the automotive, trucking, construction, mining, and agriculture sectors. Its toolchain includes synthetic data generation and data mining, scenario simulation and perception systems, motion planning and control development, sensor suite design, verification/validation, and driver data management. The funds will be used for computing infrastructure, recruitment, and expansion of its product portfolio. The company was founded in 2017 and is headquartered in Mountain View, California, United States.
*Disclaimer: This article was created by the original author. The content of the article is his personal point of view, and our reprint is only for sharing and discussion, and does not mean that we agree or agree, if you have any objections, please contact the background.