Semiconductor companies in the USA hiring now

Cerebras Systems

Sunnyvale, US · Series G

Cerebras Systems builds the world's largest AI chip, 56 times larger than GPUs. Its products include the Wafer Scale Engine 3 (WSE-3) semiconductors, CS-3 supercomputers, and AI inference and training cloud APIs. The company delivers sub-second inference speeds and breakthrough training performance for enterprises, research labs, and government agencies.

89 open roles12% remote

Etched

San Jose, US

The company co-designs chips, racks, software, and manufacturing to deliver high-throughput, low-latency inference for both prefill and decode workloads. Its products are sold as full rack-scale systems, not just chips.

87 open roles2% remote

Tenstorrent

Toronto, CA · Series D

Tenstorrent designs and manufactures AI hardware and software for data processing and machine learning tasks. Tenstorrent develops AI hardware and software solutions for data processing and machine learning application.

68 open roles31% remote

Altera

San José, CR · Private Equity

Altera’s broad end-to-end portfolio includes FPGAs, CPLDs, software, IP, development tools and accelerated computing platforms that provide the flexibility and performance needed to accelerate innovation across industries.

68 open roles37% remote

Lightmatter

Mountain View, US · Series D

Lightmatter builds photonic interconnects and AI accelerators that use light instead of electrical signals to process and transmit data. Its Passage product is a silicon photonics interconnect system that transmits data between computing components using light.

62 open roles2% remote

SiFive

santa clara, US · Series G

SiFive provides RISC-V processor designs for developers building custom chips. Its portfolio supports edge computing, machine learning and AI applications.

55 open roles

PsiQuantum

Palo Alto, US · Series E

PsiQuantum is building a commercially viable silicon photonic quantum computer using photonic qubits and leveraging mature semiconductor manufacturing processes. PsiQuantum offers a software package called Construct for companies to design algorithms that might one day run on the computer.

47 open roles100% remote

Fab2

Austin, US

Fab2 builds semiconductor fabrication equipment and software for chip design. It makes its own tools and components and uses automation to develop manufacturing processes.

47 open roles0% remote

Ayar Labs

San Jose, US · Series D

Ayar Labs develops co-packaged optics (CPO) solutions that use light instead of electricity to move data faster and with lower power. Its TeraPHY optical engine and SuperNova light source enable high-bandwidth, low-latency data transfer for AI and high-performance computing infrastructure.

37 open roles0% remote

MatX

Mountain View, US · Series B

MatX designs hardware tailored for the world's best AI models, dedicating every transistor to maximizing performance for large models. MatX makes high-throughput chips for LLMs, targeting training, RL, inference prefill, and inference decode.

36 open roles81% remote

Omni Design Technologies

Milpitas, US · Series A

Omni Design Technologies provides high-performance, ultra-low power IP solutions across advanced CMOS nodes. It delivers Wideband Signal Processing solutions using high-performance, ultra-low power semiconductor IP core products built on advanced FinFET process nodes.

30 open roles0% remote

Velaura AI

santa clara, US · Series C

Velaura AI develops ultra-low-power compute technology for cloud, edge, and physical AI applications.

29 open roles0% remote

Lumilens

San Jose, US

Lumilens designs, develops, and manufactures near-package optics, co-packaged optics, and pluggable optical transceivers for hyperscaler AI data centers. The company's LumiCore platform spans silicon photonics, mixed-signal ICs, electrical-optical interposers, and optical systems.

25 open roles0% remote

DensityAI

Mountain View, US · Seed

DensityAI develops AI chips, hardware and software for data centers.

24 open roles

Efficient Computer

Pittsburgh, US · Series A

Efficient Computer builds the world's most energy-efficient general-purpose processors, delivering up to 100× better efficiency than low-power CPUs. Its Electron E1 processor is built on the Efficient Fabric spatial dataflow architecture and is programmable using standard high-level languages and AI/ML frameworks.

23 open roles67% remote

Celero Communications

irvine, US · Series B

Optical networking chips for AI data centers

20 open roles

Mythic

Austin, US

Mythic is the only analog compute-in-memory platform delivering 100x efficiency gains for the innovators who need AI inference at scale without the unsustainable energy cost, with technology proven across data center, automotive, robotics, and defense applications.

19 open roles15% remote

Positron

reno, US · Series B

Positron develops silicon chips and accelerators for running AI models.

18 open roles72% remote

Tensordyne

Sunnyvale, US · Series C

Tensordyne provides custom silicon and systems designed to run large artificial intelligence (AI) models, aiming to reduce infrastructure and operational costs. The company offers purpose-built processors for compute intensive visual perception AI inference processing.

17 open roles

Unconventional AI

San Francisco, US · Seed

Unconventional AI is building a new physical substrate for intelligence to achieve biology-scale energy efficiency. The company offers a novel computing architecture that bypasses the limitations of traditional von Neumann systems, enabling customers to accelerate artificial intelligence workloads with energy efficiency and reduced data bottlenecks.

16 open roles100% remote

Eliyan

Santa Clara, US · Series C

Eliyan provides silicon-proven NuLink PHYs and NuGear chiplet families for die-to-die and chip-to-chip connections. Its technologies enable customers to build higher-performance systems with greater memory capacity and improved efficiency.

13 open roles0% remote

nEye Systems

Santa Clara, US · Series C

nEye develops MEMS-based silicon photonics optical circuit switches (OCS) that enable direct optical connections among thousands of GPUs and memory units. The company's SuperSwitch is an ultra-low power consumption, high radix, compact chip-scale design for hyperscale data centers.

11 open roles0% remote

HyperLight

Cambridge, US · Series C

HyperLight develops high-performance photonic integrated circuits (PICs) powered by thin-film lithium niobate (TFLN). The company offers a comprehensive suite of TFLN-based products and customized solutions spanning design, manufacturing, advanced testing, and integration.

11 open roles0% remote

TYLsemi

TYLsemi supplies chiplets for AI computing infrastructure. Its products provide connectivity, power and efficiency functions that customers integrate into advanced silicon systems.

11 open roles43% remote

NextSilicon

Givatayim, IL · Series C

NextSilicon’s flagship product is the Maverick-2 Intelligent Compute Accelerator, a dataflow accelerator that provides adaptable hardware acceleration for HPC and AI applications.

10 open roles27% remote

Extropic

Austin, US · Seed

Extropic is building thermodynamic computing hardware that is radically more energy efficient than GPUs.

9 open roles0% remote

Ethernovia

San Jose, US · Series B

Ethernovia develops Ethernet-based networking solutions for software-defined and autonomous vehicles, robotics, and intelligent machines.

9 open roles21% remote

Architect Labs

Palo Alto, US · Seed

Architect Labs builds AI models and tools for on-demand custom ASICs at scale. Its AI system co-designs, optimizes, and provably verifies chips end-to-end for a customer's specification or target workload.

8 open roles0% remote

AttoTude

Menlo Park, US · Series C

AttoTude is developing next-generation AI infrastructure and networking technologies designed to dramatically improve the reliability and economics of AI interconnect systems. AttoTude is pioneering a next-generation interconnect platform for AI and hyperscale infrastructure that combines advanced ASIC signal generation with low-loss dielectric waveguides to enable high-bandwidth connectivity.

7 open roles50% remote

PowerLattice

Vancouver, US · Series A

The chiplet tightly couples power and compute, reducing total compute power needs by more than 50%.

6 open roles0% remote