Developer docs
API playgroundTry for free, no card

Search company profiles

NextSilicon

Full company profile

uuid0000s2j

Namestring
NextSilicon
Legal namestring
Next Silicon Ltd.
Websiteurl
nextsilicon.com
Company typeenum
Private
Founded yearint
2017
Descriptiontext

NextSilicon is an Israel-based fabless semiconductor company founded in 2017 that designs and sells dataflow accelerators and RISC-V CPUs targeting high-performance computing (HPC) and AI infrastructure markets. Its core technology is the Intelligent Compute Architecture (ICA), a non-Von Neumann approach in which computation is driven by data availability rather than a fixed program counter, allowing the majority of silicon to be allocated to arithmetic logic units instead of control logic. The architecture supports runtime hardware reconfiguration based on real-time telemetry, and the programming model accepts unmodified C/C++, Python, Fortran, CUDA, and standard AI frameworks, eliminating the need for proprietary languages or code rewrites.

The company's primary product is the Maverick-2 dataflow accelerator, available as PCIe cards and OCP Accelerator Modules with up to 192GB of HBM3E memory and fabricated on TSMC 5nm. A second product, the Arbel 64-core RISC-V server-class CPU, targets data center, HPC, and AI workloads on a production roadmap for early 2026. Maverick-2 is in production as of Q4 2025, with the flagship Spectra supercomputer deployment of 128 accelerators achieving full system acceptance at Sandia National Laboratories in May 2026 under the US Department of Energy's Vanguard program for the National Nuclear Security Administration's Advanced Simulation and Computing initiative. Lawrence Livermore and Los Alamos national laboratories also participate in the program.

NextSilicon sells hardware directly to government and research institutions and reaches broader enterprise markets through OEM and integration partnerships with Penguin Solutions, Dell, Databank, and Cornelis Networks. The company is headquartered in Israel with offices across the United States, Serbia, Germany, Switzerland, India, and Australia, employing 350+ people globally. Cumulative funding exceeds $300 million from investors including Amiti, Aleph, Playground Global, Third Point Ventures, and StepStone.

Short descriptiontext

NextSilicon is an Israel-based fabless semiconductor company that designs dataflow accelerators (Maverick-2) and RISC-V CPUs (Arbel) for HPC and AI infrastructure, selling to US government national laboratories and enterprise customers via OEM partners including Penguin Solutions, Dell, and Cornelis Networks.

Operating statusenum
Operating
Ownership categoryenum
Headcount rangeband
251–500
akta.pro rankint
HeadquartersGiv'atayim, Israel
HQ citystring
Giv'atayim
HQ countrystring
Israel
HQ regionstring
Middle East
Markets served

Serves global market

Offices9 records

Each record includes

City, Country, Type, Description, Source

Keyword5 values
dataflow accelerators, HPC processors, RISC-V server CPUs, AI infrastructure chips, reconfigurable computing
Industry3 codes
PRIMARYHDAHAJAF
AI/Neural Processing Units (NPUs) & ML Accelerators
SECONDARYHDACAGAA
Graphics & Accelerator Cards (GPUs, compute accelerators)
Hardware, Technology & Telecom/Computing Hardware & Peripherals/PC Components & Upgrades (Motherboards, GPUs, RAM, PSUs, Cooling)
SECONDARYHDAAAAAI
AI Compiler, Runtime & Kernel Optimization Software (CUDA/ROCm/XLA, graph compilers)
Hardware, Technology & Telecom/Artificial Intelligence & Machine Learning/AI Infrastructure & Compute (GPUs/ASICs, accelerators)
Product category
Dataflow Accelerators / HPC Silicon
GTM motion1 record

Each record includes

Type, Description, Source

Revenue model2 records
1Maverick-2 Hardware Sales
TypeHardware Sales
Description

NextSilicon sells its Maverick-2 dataflow accelerators as PCIe cards and OCP Accelerator Modules (OAM) to enterprise and government customers. Available in production volumes as of Q4 2025. Customers include national laboratories, HPC research institutions, and AI infrastructure operators.

nextsilicon.com
2Arbel RISC-V CPU Sales
TypeHardware Sales
Description

Enterprise-grade 64-core and 128-core RISC-V server processors sold to data center, HPC, and AI infrastructure customers. Currently in production with customer and partner silicon evaluations underway.

nextsilicon.com
Marketing channels6 records

Each record includes

Title, Type, Stage, Description, Source

Distribution channels5 records

Each record includes

Title, Type, Scope, Target buyer, Description, Source

Cost components5 values
Technology or R&D, Supply Chain, Personnel, Marketing or Sales, Operations
GTM typeB2B
B2B
Offering typeHardware or Manufacturing
Hardware or Manufacturing
Brand1 of 4 records shown
1Maverick-2
Description

First production dataflow accelerator powered by NextSilicon's Intelligent Compute Architecture (ICA), delivering up to 10x performance at less than half the power of leading GPUs for HPC and AI workloads.

nextsilicon.com
+3 more records
Core offering1 text field

NextSilicon designs, manufactures, and sells the Maverick-2 dataflow accelerator and the Arbel RISC-V server CPU, both built on its proprietary Intelligent Compute Architecture (ICA). Maverick-2 is sold as PCIe cards and OCP Accelerator Modules (OAM) to HPC and AI infrastructure customers, delivering runtime-reconfigurable hardware that adapts to software in real time. The products enable unmodified legacy code (CUDA, C/C++, Python, Fortran) to run without porting, targeting government national laboratories, research institutions, and AI infrastructure operators.

Differentiator
Functional benefit
Problem solved
Quantifiable outcome1 of 3 values shown
  • Up to 10x performance over leading GPUs at 60% less power in algorithmically complex workloads; 600 GFLOPS HPCG at 750W (matching GPUs at half power); 32.6 GUPS at 460W (22x faster than CPUs, ~6x faster than GPUs); PageRank 10x higher graph analytics performance than leading GPUs on 25GB+ graph sizes that GPUs cannot process
+2 more records
Product overview1 text field

NextSilicon's product portfolio centers on its Intelligent Compute Architecture (ICA) platform, which powers two primary products: Maverick-2, a runtime-reconfigurable dataflow accelerator for HPC and AI workloads available in PCIe (single-die, 400W) and OAM (dual-die, 750W) form factors; and Arbel, a 64-core RISC-V server-class CPU designed as a companion processor to Maverick-2 for data center and HPC environments. Both products leverage ICA's adaptive software intelligence to dynamically optimize hardware configurations based on workload behavior.

Product and service2 records
1Maverick-2
CategoryDataflow Accelerator Hardware
Description

Maverick-2 is a runtime-reconfigurable dataflow accelerator powered by NextSilicon's Intelligent Compute Architecture (ICA). It is sold as PCIe cards (single-die, 96GB HBM3E, 400W TDP) and OCP Accelerator Modules (OAM, dual-die, 192GB HBM3E, 750W TDP) to HPC and AI infrastructure customers. It features 54 billion transistors fabricated at TSMC 5nm and supports unmodified C/C++, Python, Fortran, CUDA, Kokkos, ROCm/HIP, OpenCL, TensorFlow, and OneAPI code without porting.

2Arbel
CategoryRISC-V Server Processor Hardware
Description

Arbel is a 64-core server-class RISC-V processor (RVA23 Hypervisor) with a 10-wide instruction pipeline, 3.4 GHz target frequency, three 256-bit vector units, and 480-entry reorder buffer. Built on TSMC 5nm process, it is designed as a companion CPU to Maverick-2 for data center, HPC, and AI workloads. Designed to match or beat x86 incumbents on SPECint and SPECfp benchmarks per GHz.

Scale indicator7 records

Each record includes

Type, Value, Description, Source

Partnership7 partners
Strategic tierCoreTypeStrategic or Co-development PartnerAnnounced on2026-06-23
Description

Cornelis Networks and NextSilicon formed a partnership to jointly create AI and high-performance computing reference architectures combining Cornelis' CN5000 fabric with NextSilicon's Maverick-2 compute platform. The converged stack aims to eliminate infrastructure bottlenecks (network congestion causing compute idle time and von Neumann architecture limitations) with validated system blueprints to be commercialized through OEM partners. Companies plan to extend testing to 800 Gbps CN6000 fabric in H2 2026 and explore disaggregated inference and agentic AI workloads.

Strategic tierFlagshipTypeStrategic or Co-development PartnerAnnounced on2026-05-18
Description

Sandia National Laboratories, as part of a tri-lab consortium with Lawrence Livermore and Los Alamos National Laboratories, partnered with NextSilicon under the Vanguard program to deploy the Spectra supercomputer — the first system to incorporate NextSilicon's Maverick-2 architecture. Full system acceptance was granted in May 2026, qualifying Maverick-2 for nuclear safety-related computational testing. The 64-node machine with 128 Maverick-2 dual-die accelerators runs workloads under NNSA's Advanced Simulation and Computing program for modeling nuclear reactions.

Strategic tierCoreTypeImplementation/ SI/ Consulting PartnerAnnounced on2025-10-22
Description

Penguin Solutions served as the system integrator for building and deploying the Spectra supercomputer at Sandia National Laboratories with NextSilicon's Maverick-2 chips. The collaboration enabled the 64-node, 128-accelerator system deployment and represents Penguin's role in bringing NextSilicon's technology to government HPC environments.

Strategic tierCoreTypeStrategic or Co-development PartnerAnnounced on2023-01-01
Description

Lawrence Livermore National Laboratory (LLNL) is part of the tri-lab consortium with Sandia and Los Alamos, leading a partnership with NextSilicon to deliver the Advanced Architecture Prototype System (AAPS) under Sandia's Vanguard program. The tri-lab consortium evaluates emerging pre-tested technologies for their viability in supporting the NNSA's Advanced Simulation and Computing (ASC) program.

5Los Alamos National Laboratory
Strategic tierCoreTypeStrategic or Co-development PartnerAnnounced on2023-01-01
Description

Los Alamos National Laboratory (LANL) is part of the tri-lab consortium with Sandia and Lawrence Livermore, leading a partnership with NextSilicon to deliver the Advanced Architecture Prototype System (AAPS) under Sandia's Vanguard program. The tri-lab consortium evaluates emerging pre-tested technologies for their viability in supporting the NNSA's ASC program.

nextsilicon.com
Strategic tierMinorTypeChannel Partner/ Reseller/ Distributor
Description

Dell Technologies is listed as part of NextSilicon's rapidly expanding partner ecosystem with native support for commercial HPC libraries. Dell provides HPC infrastructure distribution capabilities.

Strategic tierMinorTypeChannel Partner/ Reseller/ Distributor
Description

Databank is listed as part of NextSilicon's partner ecosystem providing HPC infrastructure distribution capabilities.

Recent move6 records

Each record includes

Date, Type, Title, Description, Source

Expansion highlight5 records

Each record includes

Type, Description

Peers10 records
TypeDirect peer
Description

Dataflow AI accelerator (RDU, SN40L) targeting enterprise and government AI deployments with a reconfigurable dataflow architecture conceptually similar to NextSilicon's ICA — both attack the Von Neumann bottleneck and target national-lab/government HPC and AI workloads.

TypeDirect peer
Description

AI inference accelerator (LPU) using a deterministic dataflow architecture; targets government (including US Air Force) and enterprise inference workloads with low-latency claims. Closely comparable in alternative-architecture positioning and GTM toward federal AI infrastructure buyers.

3Tenstorrent
TypeDirect peer
Description

AI accelerator company building RISC-V-based Wormhole/Blackhole chips for AI training and inference. Direct comparable on the RISC-V + AI silicon stack and overlaps with NextSilicon's combined Maverick-2 + Arbel proposition.

TypeDirect peer
Description

AI accelerator company selling wafer-scale CS-3 chips to government labs (including Argonne, LLNL) and enterprise AI customers. Closest direct competitor in the 'alternative architecture to GPU' positioning NextSilicon pursues, with similar national-lab GTM.

TypeDirect peer
Description

Leading commercial RISC-V CPU core IP and developer of high-performance RISC-V application processors. Direct comparable to Arbel on the RISC-V server CPU side, with overlapping customer interest from hyperscalers and AI infrastructure builders.

6Rivos
TypeDirect peer
Description

Stealth-stage RISC-V server CPU startup targeting data-center and HPC workloads with high-performance RISC-V cores. Closest head-to-head competitor to Arbel for RISC-V server CPU share and shares the same ecosystem tailwinds around RISC-V adoption.

TypeBroad incumbent
Description

Supplier of Instinct MI300X/MI325X/MI350 accelerators and EPYC CPUs for HPC and AI. Competes head-to-head with NextSilicon in FP64 and HPC accelerator slots at national labs and competes with Arbel in the server-CPU market.

TypeBroad incumbent
Description

Dominant supplier of GPUs (H100, Blackwell B200, forthcoming Rubin) for AI training/inference and HPC. Direct competitor to Maverick-2 in accelerator silicon and the benchmark against which NextSilicon's perf/W claims are measured; CUDA software ecosystem is the central competitive moat NextSilicon explicitly targets.

TypeBroad incumbent
Description

Intel's accelerator unit selling Gaudi 2/3 and the now-discontinued Ponte Vecchio GPUs for AI training/inference and HPC. Competes with NextSilicon in HPC and AI accelerator procurement at large enterprises and government buyers, while leveraging Intel's broader x86 server stack.

TypeDirect peer
Description

Bow-class IPU accelerator targeting HPC and AI workloads with a non-GPU architecture (originally backed by Sequoia, now folded into SoftBank). Comparable as a struggling dataflow/parallel-processing alternative whose trajectory illustrates the difficulty of scaling versus NVIDIA.

Market position
Strengths5 records

Each record includes

Headline, Details, Source

Weaknesses5 records

Each record includes

Headline, Details, Source

Competitive moat5 records

Each record includes

Type, Details

Key risks6 records

Each record includes

Headline, Details, Source

Key highlights7 records

Each record includes

Headline, Details, Source

Customer concentration

Classification, Details

Named customers4 records

Each record includes

Name, Industry, Type, Use case, Source, UUID

Segment3 records

Each record includes

Title, Type, Primary, Description, Pain point addressed, Use case, Source

Ideal customer profile3 records

Each record includes

Profile, Firmographic size, Sales motion, Sales cycle length, Buying structure, Purchase trigger, Buyer persona, Geography, Industry vertical, Primary use case, Description, Pain points, Evidence proof points, Target buyer

Technology focused
Yes
API detail
Has APIbool
No

Docs URL, Description

AI capability3 records

Each record includes

Type, Description, Source

AI maturity
App detail

Has app

Feature5 records

Each record includes

Title, Differentiator, Description, Source

Core technology
Revenue estimate
Valuation estimate
Number of profiles
Profiles2 records

Each record includes

Name, Designation, Designation category, Overview, Profile commentary, Source

No data
No data
Funding overview

Funding stage, Last funding date, Total funding USD

Funding rounds5 records

Each record includes

Round, Amount USD, Date, Pre money valuation, Total investors, Investors, News

Investors11 records

Each record includes

Name, Type, Date of entry, Rounds participated, Website

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Each record includes

Name, Acquisition type, Announced date, Completed date, Status, Website, News

Investment

Each record includes

Name, Round, Announced date, Lead investor, Website, News

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

NextSilicon

Dataflow Accelerators / HPC Siliconnextsilicon.com

NextSilicon is an Israel-based fabless semiconductor company that designs dataflow accelerators (Maverick-2) and RISC-V CPUs (Arbel) for HPC and AI infrastructure, selling to US government national laboratories and enterprise customers via OEM partners including Penguin Solutions, Dell, and Cornelis Networks.

What NextSilicon does

NextSilicon is an Israel-based fabless semiconductor company founded in 2017 that designs and sells dataflow accelerators and RISC-V CPUs targeting high-performance computing (HPC) and AI infrastructure markets. Its core technology is the Intelligent Compute Architecture (ICA), a non-Von Neumann approach in which computation is driven by data availability rather than a fixed program counter, allowing the majority of silicon to be allocated to arithmetic logic units instead of control logic. The architecture supports runtime hardware reconfiguration based on real-time telemetry, and the programming model accepts unmodified C/C++, Python, Fortran, CUDA, and standard AI frameworks, eliminating the need for proprietary languages or code rewrites.

The company's primary product is the Maverick-2 dataflow accelerator, available as PCIe cards and OCP Accelerator Modules with up to 192GB of HBM3E memory and fabricated on TSMC 5nm. A second product, the Arbel 64-core RISC-V server-class CPU, targets data center, HPC, and AI workloads on a production roadmap for early 2026. Maverick-2 is in production as of Q4 2025, with the flagship Spectra supercomputer deployment of 128 accelerators achieving full system acceptance at Sandia National Laboratories in May 2026 under the US Department of Energy's Vanguard program for the National Nuclear Security Administration's Advanced Simulation and Computing initiative. Lawrence Livermore and Los Alamos national laboratories also participate in the program.

NextSilicon sells hardware directly to government and research institutions and reaches broader enterprise markets through OEM and integration partnerships with Penguin Solutions, Dell, Databank, and Cornelis Networks. The company is headquartered in Israel with offices across the United States, Serbia, Germany, Switzerland, India, and Australia, employing 350+ people globally. Cumulative funding exceeds $300 million from investors including Amiti, Aleph, Playground Global, Third Point Ventures, and StepStone.

NextSilicon firmographics

Firmographics
Name
NextSilicon
Legal name
Next Silicon Ltd.
Website
https://nextsilicon.com
Company type
Private
Founded year
2017
Operating status
Operating
Headcount range
251–500 employees
Short description
NextSilicon is an Israel-based fabless semiconductor company that designs dataflow accelerators (Maverick-2) and RISC-V CPUs (Arbel) for HPC and AI infrastructure, selling to US government national laboratories and enterprise customers via OEM partners including Penguin Solutions, Dell, and Cornelis Networks.
Ownership category
akta.pro rank

NextSilicon industry classification

Industry
Product category
Dataflow Accelerators / HPC Silicon
akta.pro industry
PRIMARYHDAHAJAF
AI/Neural Processing Units (NPUs) & ML Accelerators
SECONDARYHDACAGAA
Graphics & Accelerator Cards (GPUs, compute accelerators)
Hardware, Technology & Telecom/Computing Hardware & Peripherals/PC Components & Upgrades (Motherboards, GPUs, RAM, PSUs, Cooling)
SECONDARYHDAAAAAI
AI Compiler, Runtime & Kernel Optimization Software (CUDA/ROCm/XLA, graph compilers)
Hardware, Technology & Telecom/Artificial Intelligence & Machine Learning/AI Infrastructure & Compute (GPUs/ASICs, accelerators)

Keywords

  • Dataflow accelerators
  • HPC processors
  • RISC-V server CPUs
  • AI infrastructure chips
  • Reconfigurable computing

Where NextSilicon is headquartered

Location

Headquarters

HQ city
Giv'atayim
HQ country
Israel
HQ region
Middle East

Offices9 records

Markets served

NextSilicon business model

Business model
GTM type
B2B
Offering type
Hardware or Manufacturing
Cost components
Technology or R&D, Supply Chain, Personnel, Marketing or Sales, Operations

Revenue model

  1. Maverick-2 Hardware Sales: NextSilicon sells its Maverick-2 dataflow accelerators as PCIe cards and OCP Accelerator Modules (OAM) to enterprise and government customers. Available in production volumes as of Q4 2025. Customers include national laboratories, HPC research institutions, and AI infrastructure operators.
  2. Arbel RISC-V CPU Sales: Enterprise-grade 64-core and 128-core RISC-V server processors sold to data center, HPC, and AI infrastructure customers. Currently in production with customer and partner silicon evaluations underway.

Go-to-market motion1 record

Distribution channels5 records

Marketing channels6 records

NextSilicon product offering

Product offering

Core offering

NextSilicon designs, manufactures, and sells the Maverick-2 dataflow accelerator and the Arbel RISC-V server CPU, both built on its proprietary Intelligent Compute Architecture (ICA). Maverick-2 is sold as PCIe cards and OCP Accelerator Modules (OAM) to HPC and AI infrastructure customers, delivering runtime-reconfigurable hardware that adapts to software in real time. The products enable unmodified legacy code (CUDA, C/C++, Python, Fortran) to run without porting, targeting government national laboratories, research institutions, and AI infrastructure operators.

Product overview

NextSilicon's product portfolio centers on its Intelligent Compute Architecture (ICA) platform, which powers two primary products: Maverick-2, a runtime-reconfigurable dataflow accelerator for HPC and AI workloads available in PCIe (single-die, 400W) and OAM (dual-die, 750W) form factors; and Arbel, a 64-core RISC-V server-class CPU designed as a companion processor to Maverick-2 for data center and HPC environments. Both products leverage ICA's adaptive software intelligence to dynamically optimize hardware configurations based on workload behavior.

Differentiator

Problem solved

Functional benefit

Brands

  • Maverick-2: First production dataflow accelerator powered by NextSilicon's Intelligent Compute Architecture (ICA), delivering up to 10x performance at less than half the power of leading GPUs for HPC and AI workloads.
  • Arbel
  • Intelligent Compute Architecture (ICA)
  • Spectra

Products and services

  • Maverick-2 Maverick-2 is a runtime-reconfigurable dataflow accelerator powered by NextSilicon's Intelligent Compute Architecture (ICA). It is sold as PCIe cards (single-die, 96GB HBM3E, 400W TDP) and OCP Accelerator Modules (OAM, dual-die, 192GB HBM3E, 750W TDP) to HPC and AI infrastructure customers. It features 54 billion transistors fabricated at TSMC 5nm and supports unmodified C/C++, Python, Fortran, CUDA, Kokkos, ROCm/HIP, OpenCL, TensorFlow, and OneAPI code without porting.
  • Arbel Arbel is a 64-core server-class RISC-V processor (RVA23 Hypervisor) with a 10-wide instruction pipeline, 3.4 GHz target frequency, three 256-bit vector units, and 480-entry reorder buffer. Built on TSMC 5nm process, it is designed as a companion CPU to Maverick-2 for data center, HPC, and AI workloads. Designed to match or beat x86 incumbents on SPECint and SPECfp benchmarks per GHz.

Quantifiable outcome

  • Up to 10x performance over leading GPUs at 60% less power in algorithmically complex workloads; 600 GFLOPS HPCG at 750W (matching GPUs at half power); 32.6 GUPS at 460W (22x faster than CPUs, ~6x faster than GPUs); PageRank 10x higher graph analytics performance than leading GPUs on 25GB+ graph sizes that GPUs cannot process
  • +2 more outcomes

Companies that use NextSilicon

Customer profile

Named customers4 records

Segments3 records

Ideal customer profiles3 records

NextSilicon technology and API

Technology

Technology focussed Yes

API detail

Has API
No
API docs
API detail

Core technology

AI maturity

App detail

AI capability3 records

Feature5 records

NextSilicon partnerships and signals

Strategic signal

Partnerships

Seven partnerships are on record, tiered core, flagship and minor.

  • Cornelis NetworkscoreStrategic or Co-development Partner · 23 June 2026Cornelis Networks and NextSilicon formed a partnership to jointly create AI and high-performance computing reference architectures combining Cornelis' CN5000 fabric with NextSilicon's Maverick-2 compute platform. The converged stack aims to eliminate infrastructure bottlenecks (network congestion causing compute idle time and von Neumann architecture limitations) with validated system blueprints to be commercialized through OEM partners. Companies plan to extend testing to 800 Gbps CN6000 fabric in H2 2026 and explore disaggregated inference and agentic AI workloads.
  • Sandia National LaboratoriesflagshipStrategic or Co-development Partner · 18 May 2026Sandia National Laboratories, as part of a tri-lab consortium with Lawrence Livermore and Los Alamos National Laboratories, partnered with NextSilicon under the Vanguard program to deploy the Spectra supercomputer — the first system to incorporate NextSilicon's Maverick-2 architecture. Full system acceptance was granted in May 2026, qualifying Maverick-2 for nuclear safety-related computational testing. The 64-node machine with 128 Maverick-2 dual-die accelerators runs workloads under NNSA's Advanced Simulation and Computing program for modeling nuclear reactions.
  • Penguin SolutionscoreImplementation/ SI/ Consulting Partner · 22 October 2025Penguin Solutions served as the system integrator for building and deploying the Spectra supercomputer at Sandia National Laboratories with NextSilicon's Maverick-2 chips. The collaboration enabled the 64-node, 128-accelerator system deployment and represents Penguin's role in bringing NextSilicon's technology to government HPC environments.
  • Lawrence Livermore National LaboratorycoreStrategic or Co-development Partner · 1 January 2023Lawrence Livermore National Laboratory (LLNL) is part of the tri-lab consortium with Sandia and Los Alamos, leading a partnership with NextSilicon to deliver the Advanced Architecture Prototype System (AAPS) under Sandia's Vanguard program. The tri-lab consortium evaluates emerging pre-tested technologies for their viability in supporting the NNSA's Advanced Simulation and Computing (ASC) program.
  • Los Alamos National LaboratorycoreStrategic or Co-development Partner · 1 January 2023Los Alamos National Laboratory (LANL) is part of the tri-lab consortium with Sandia and Lawrence Livermore, leading a partnership with NextSilicon to deliver the Advanced Architecture Prototype System (AAPS) under Sandia's Vanguard program. The tri-lab consortium evaluates emerging pre-tested technologies for their viability in supporting the NNSA's ASC program.
  • Dell TechnologiesminorChannel Partner/ Reseller/ DistributorDell Technologies is listed as part of NextSilicon's rapidly expanding partner ecosystem with native support for commercial HPC libraries. Dell provides HPC infrastructure distribution capabilities.
  • DatabankminorChannel Partner/ Reseller/ DistributorDatabank is listed as part of NextSilicon's partner ecosystem providing HPC infrastructure distribution capabilities.

Scale indicators7 records

Recent moves6 records

Expansion highlights5 records

NextSilicon competitors and assessment

Company assessment

Direct peers

  • SambaNova Systems: Dataflow AI accelerator (RDU, SN40L) targeting enterprise and government AI deployments with a reconfigurable dataflow architecture conceptually similar to NextSilicon's ICA — both attack the Von Neumann bottleneck and target national-lab/government HPC and AI workloads.
  • Groq: AI inference accelerator (LPU) using a deterministic dataflow architecture; targets government (including US Air Force) and enterprise inference workloads with low-latency claims. Closely comparable in alternative-architecture positioning and GTM toward federal AI infrastructure buyers.
  • Tenstorrent: AI accelerator company building RISC-V-based Wormhole/Blackhole chips for AI training and inference. Direct comparable on the RISC-V + AI silicon stack and overlaps with NextSilicon's combined Maverick-2 + Arbel proposition.
  • Cerebras Systems: AI accelerator company selling wafer-scale CS-3 chips to government labs (including Argonne, LLNL) and enterprise AI customers. Closest direct competitor in the 'alternative architecture to GPU' positioning NextSilicon pursues, with similar national-lab GTM.
  • SiFive: Leading commercial RISC-V CPU core IP and developer of high-performance RISC-V application processors. Direct comparable to Arbel on the RISC-V server CPU side, with overlapping customer interest from hyperscalers and AI infrastructure builders.
  • Rivos: Stealth-stage RISC-V server CPU startup targeting data-center and HPC workloads with high-performance RISC-V cores. Closest head-to-head competitor to Arbel for RISC-V server CPU share and shares the same ecosystem tailwinds around RISC-V adoption.
  • Graphcore: Bow-class IPU accelerator targeting HPC and AI workloads with a non-GPU architecture (originally backed by Sequoia, now folded into SoftBank). Comparable as a struggling dataflow/parallel-processing alternative whose trajectory illustrates the difficulty of scaling versus NVIDIA.

Broad incumbents

  • AMD: Supplier of Instinct MI300X/MI325X/MI350 accelerators and EPYC CPUs for HPC and AI. Competes head-to-head with NextSilicon in FP64 and HPC accelerator slots at national labs and competes with Arbel in the server-CPU market.
  • NVIDIA: Dominant supplier of GPUs (H100, Blackwell B200, forthcoming Rubin) for AI training/inference and HPC. Direct competitor to Maverick-2 in accelerator silicon and the benchmark against which NextSilicon's perf/W claims are measured; CUDA software ecosystem is the central competitive moat NextSilicon explicitly targets.
  • Intel (Habana Labs): Intel's accelerator unit selling Gaudi 2/3 and the now-discontinued Ponte Vecchio GPUs for AI training/inference and HPC. Competes with NextSilicon in HPC and AI accelerator procurement at large enterprises and government buyers, while leveraging Intel's broader x86 server stack.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights7 records

Customer concentration

NextSilicon social profiles

Digital presence

NextSilicon financial estimates

Financial estimate

Revenue estimate

Valuation estimate

NextSilicon leadership team

Management profile

Number of profiles

Profiles2 records

NextSilicon funding detail

Funding detail

Funding overview

Funding rounds5 records

Investors11 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

NextSilicon M&A and investment

M&A and investment

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about NextSilicon

What does NextSilicon do?

NextSilicon designs, manufactures, and sells the Maverick-2 dataflow accelerator and the Arbel RISC-V server CPU, both built on its proprietary Intelligent Compute Architecture (ICA). Maverick-2 is sold as PCIe cards and OCP Accelerator Modules (OAM) to HPC and AI infrastructure customers, delivering runtime-reconfigurable hardware that adapts to software in real time. The products enable unmodified legacy code (CUDA, C/C++, Python, Fortran) to run without porting, targeting government national laboratories, research institutions, and AI infrastructure operators.

Is NextSilicon a public or private company?

NextSilicon is a private company. It is classified as venture growth investor backed and is currently operating.

When was NextSilicon founded?

NextSilicon was founded in 2017. It employs 251 to 500 people.

Where is NextSilicon based?

NextSilicon is headquartered in Giv'atayim, Israel, in the Middle East region.

How does NextSilicon make money?

Two revenue lines are on record. Maverick-2 Hardware Sales are the primary driver. The others are arbel RISC-V CPU Sales.

Who are NextSilicon's main competitors?

Direct peers on record are SambaNova Systems, Groq, Tenstorrent, Cerebras Systems, SiFive, Rivos and Graphcore. Broad incumbents are AMD, NVIDIA and Intel (Habana Labs).

Does NextSilicon have an API?

No public API is recorded for NextSilicon.

What industry is NextSilicon in?

NextSilicon's product category is Dataflow Accelerators / HPC Silicon. Its primary akta.pro industry code is HDAHAJAF, AI/Neural Processing Units (NPUs) & ML Accelerators, with a secondary code of HDACAGAA, Graphics & Accelerator Cards (GPUs, compute accelerators).

Unlock the full company data

50 free credits on sign-up, no credit card required.

Contact sales
Live signals
citybizNextSilicon Appoints Kit Chee Chief Sales Officer and Kristin Owens Chief Marketing OfficerNextSilicon appointed Kit Chee as chief sales officer and Kristin Owens as chief marketing officer. Both will report to Rupal Hollenbeck, president and chief business officer, as the company expands deployments of its Maverick-2 accelerator platform for AI and high-performance computing.Pulse 2.0Lightspeed Adds NextSilicon President Rupal Hollenbeck To Board As Company Targets Profitable GrowthLightspeed Commerce appointed Rupal Hollenbeck, President and Chief Business Officer of NextSilicon, to its Board of Directors, effective October 1, 2026. She previously led go-to-market operations at Check Point, Cerebras and Oracle, and served at Intel. The appointment follows Nathalie Gaveau's departure on September 30.NewswireLightspeed Commerce Appoints Technology Industry Veteran Rupal Hollenbeck to Board of DirectorsLightspeed Commerce appointed Rupal Hollenbeck to its board of directors, effective October 1, 2026, bringing 30 years of technology leadership. She will serve as President and Chief Business Officer of NextSilicon. Nathalie Gaveau will step down from the board, effective September 30, 2026.SalesTech StarLightspeed Commerce Appoints Technology Industry Veteran Rupal Hollenbeck to Board of DirectorsLightspeed Commerce appointed Rupal Hollenbeck to its Board of Directors, effective October 1, 2026, bringing 30 years of technology and go-to-market leadership. She will serve as President and Chief Business Officer of NextSilicon. Nathalie Gaveau will step down from the board, effective September 30, 2026.FinancialContent Business PageNextSilicon Appoints Rupal Hollenbeck President and Chief Business Officer to Lead Global Commercial StrategyNextSilicon appointed Rupal Hollenbeck as President and Chief Business Officer to lead its global commercial strategy, including sales, marketing, and partnerships. This leadership change supports the company's efforts to scale adoption of its Maverick Intelligent Compute Architecture and advance its Arbel RISC-V processor roadmap. The appointment aims to strengthen commercial execution as NextSilicon expands from HPC into broader AI infrastructure markets.MorningstarNextSilicon Appoints Rupal Hollenbeck President and Chief Business Officer to Lead Global Commercial StrategyNextSilicon appointed Rupal Hollenbeck as President and Chief Business Officer to lead global commercial strategy. She brings over 30 years of tech industry experience, including roles at Check Point, Cerebras, Oracle, and Intel. The appointment supports scaling Maverick-2 and advancing Arbel RISC-V processor roadmap.HpcwireSince 1987 – Covering the Fastest Computers in the World and the People Who Run ThemA Birds of Feather session at the ISC 2026 conference brought together RISC-V proponents who discussed the open processor architecture's potential in HPC, noting over 30 billion RISC-V cores are already deployed globally. Companies including NextSilicon and Inspire Semiconductor outlined product roadmaps, with Inspire Semi planning to ship its Thunderbird board with 6,144 RISC-V cores by year-end. Participants argued RISC-V offers advantages over proprietary architectures like X86 and Arm through its open, consortium-led model, though they acknowledged the technology still needs a flagship deployment to prove its viability in HPC at scale.SDxCentralCornelis Networks, NextSilicon partner to accelerate AI and HPC infrastructureCornelis Networks and chip startup NextSilicon have formed a partnership to jointly create AI and high-performance computing reference architectures combining Cornelis' CN5000 fabric with NextSilicon's Maverick-2 compute platform. The converged stack aims to eliminate infrastructure bottlenecks that cause expensive AI and HPC systems to sit idle, with the first phase validating combined performance for OEM partners before expanding to 800 Gb/s networks. Cornelis also announced its CN5000 technology was deployed in the Stampede3 supercomputer at the Texas Advanced Computing Center, covering over 600 compute nodes for simulation workloads.MorningstarCornelis and NextSilicon to Build Joint Reference Architectures for AI and HPCCornelis and NextSilicon announced a collaboration to build joint reference architectures for AI and HPC, pairing the CN5000 fabric with the Maverick-2 accelerator. The effort targets network congestion and compute inefficiency, with evaluations underway and future testing planned on the CN6000 fabric. The companies aim to support disaggregated inference and agentic AI workloads.Business Wire BlogCornelis and NextSilicon to Build Joint Reference Architectures for AI and HPCCornelis and NextSilicon announced at ISC High Performance 2026 a collaboration to develop joint reference architectures for AI and HPC, pairing the CN5000 fabric with the Maverick-2 compute platform. The joint evaluation, already underway, aims to create validated system blueprints for OEM partners by combining congestion-free networking with workload-adaptive compute to address infrastructure bottlenecks that leave expensive AI and HPC systems underutilized. The companies plan to extend testing to the 800 Gbps CN6000 fabric in H2 2026 and explore emerging workloads including disaggregated inference and agentic AI.