Graphcore
- Company typePrivate
- Founded2016
- HeadquartersBristol, United Kingdom
- Headcount501–1,000
- GTM typeB2B
- OfferingHardware or Manufacturing
Graphcore firmographics
Firmographics- Name
- Graphcore
- Legal name
- Graphcore Limited
- Website
- https://graphcore.ai
- Company type
- Private
- Founded year
- 2016
- Operating status
- Acquired
- Headcount range
- 501–1,000 employees
- Ownership category
- akta.pro rank
Graphcore industry classification
Industry- Product category
- AI accelerator hardware
- NAICS
- Electronic Computer Manufacturing (334111), Computer Systems Design and Related Services (54151), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (5182)
- SIC
- Electronic Computers (3571), Services-Computer Integrated Systems Design (7373)
- akta.pro primary industry
- AI/Neural Processing Units (NPUs) & ML Accelerators (HDAHAJAF)
- akta.pro secondary industries
- Compute & Acceleration ICs (CPU/GPU/AI/FPGA/ASIC) (HDAHABAC), AI Server Systems & HGX/Accelerator Platforms (HDAAAAAB)
Keywords
Where Graphcore is headquartered
LocationHeadquarters
- HQ city
- Bristol
- HQ country
- United Kingdom
- HQ region
- Europe
Offices5 records
Markets served
Graphcore business model
Business model- GTM type
- B2B
- Offering type
- Hardware or Manufacturing
- Cost components
- Technology or R&D, Personnel, Supply Chain, Infrastructure, Marketing or Sales, Operations
Revenue model
- Hardware sales — IPU systems: Sale of IPU-based computing systems (Bow Pod systems, IPU-POD systems, IPU-M2000, IPU-Machines, C600 PCIe cards) integrated into computing cards, servers, and racks and manufactured/marketed by or for Graphcore.
- Software licensing — Poplar SDK and developer tools: Personal, non-exclusive, non-transferable license of the Poplar SDK, developer tools, libraries, runtimes, drivers, and Distributable Components to licensees for use in developing, testing, verifying and executing programs on Graphcore Devices, plus Container License terms for Poplar Docker containers.
- IP licensing / royalties: As part of SoftBank Group's semiconductor portfolio, Graphcore contributes IPUs, IPU-Machines, and related semiconductor technologies to SoftBank's broader IP and chip portfolio targeting Artificial Super Intelligence.
- Cloud / managed service offering via partners: IPUs are made available to users through cloud partners (Gcore Cloud, Graphcloud) and through Kubernetes-managed V-IPU partitions, supporting cloud-style consumption of IPU hardware.
- Professional services — support, training, evaluation: Discretionary support, training, and other services provided on separate pricing terms, plus a defined Evaluation License path for non-production internal business evaluation in connection with potential purchase of a Graphcore Device.
Pricing tiers
| Model | Billing | Price |
|---|---|---|
| Other | Multi-year contract | Quote-based enterprise pricing for IPU hardware and Poplar SDK licenses |
| Other | Pay-as-you-go | Evaluation License (free, time-limited) |
Go-to-market motion5 records
Distribution channels4 records
Marketing channels9 records
Graphcore product offering
Product offeringCore offering
Graphcore designs and manufactures Intelligence Processing Units (IPUs) — purpose-built AI processors with an in-processor-memory architecture — and sells them as IPU-M2000 building blocks, Bow Pod and IPU-POD scale-out systems, and C600 PCIe accelerator cards. The hardware is paired with the Poplar SDK (PopART, PopTorch, PopLibs, PopRun, PopDist) and Graphcloud, forming an integrated platform positioned as an alternative to Nvidia GPU/CUDA stacks for AI training and inference workloads. Customers include enterprises, cloud providers, AI researchers, and edge AI builders.
Product overview
Graphcore offers a unified AI compute platform that pairs its proprietary IPU (Intelligence Processing Unit) hardware with a complete software stack, rather than a single monolithic product. The core hardware lineup comprises the IPU-M2000 building block, Bow Pod and IPU-POD scale-out systems, and C600 PCIe accelerator cards, all powered by Graphcore's IPU processor based on Colossus technology. On top of the hardware, the Poplar SDK acts as the central software platform and bundles PopART (graph compiler/runtime), PopTorch (PyTorch interface), PopLibs (compute libraries), PopRun (runtime), and PopDist (distributed compute). Add-on modules extend the platform's capabilities: PopVision Tools for profiling and debugging, V-IPU for compute virtualization (exposed via the V-IPU API), and Graphcloud for managed cloud access. Together, these named products form an end-to-end stack for AI training and inference, supported by integrations with Kubernetes, Slurm, PyTorch, TensorFlow, and Docker.
Differentiator
Problem solved
Functional benefit
Brands
- Poplar SDK: Graphcore's software development kit for programming IPUs.
- PopVision Tools
- Bow Pod
- IPU-Machine
- IPU-POD
- C600
- Graphcloud
- PopART
- PopTorch
- Virtual-IPU (V-IPU)
Products and services
- IPU-M2000
- Bow Pod
- IPU-POD
- C600 PCIe Card
- Poplar SDK
- Graphcloud
Quantifiable outcome
- Raises $222M Series E (Dec 2021), bringing cumulative funding to over $710M and valuation to $2.77B
- +5 more outcomes
Companies that use Graphcore
Customer profileSegments5 records
Ideal customer profiles5 records
Graphcore technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- Yes
- API docs
- API detail
Core technology
AI maturity
App detail
Integration5 records
AI capability10 records
Feature8 records
Graphcore partnerships and signals
Strategic signalPartnerships
Seven partnerships are on record, tiered minor, core and flagship.
- Startup Policy Forum / Cambridge India Business DialogueminorVisited by Startup Policy Forum's UK Startup Safari delegation during the Cambridge India Business Dialogue, alongside other frontier UK AI firms (Beamery, Brahma AI, Nothing), to explore India-UK AI collaboration following the landmark India-UK Free Trade Agreement.
- Linaro / Arm CoreCollective consortiumcoreGraphcore joined CoreCollective, an open industry consortium led by Linaro and backed by Arm that drives collaboration and standardization across the Arm software ecosystem. The consortium also includes AMD, Ampere, Canonical, Fujitsu, Google, Microsoft, Qualcomm, Red Hat, Samsung, and SUSE, and establishes working groups covering Android, Data Center, Confidential Compute, Edge, Linux Fundamentals, Virtualization, and Windows on Arm.
- Government of India (UK-India trade mission)flagshipAnnounced the £1 billion / ~$1.3 billion Bengaluru AI Engineering Campus as part of UK PM Keir Starmer's visit to India, with support from India's 760 billion-rupee semiconductor fund aimed at attracting international chipmakers.
- Gcore CloudcoreListed as a cloud partner in the Graphcore documentation portal, with quick-start guides and partner-specific documentation for running IPUs in the Gcore Cloud environment.
- GraphcloudcoreListed as a cloud partner in the Graphcore documentation portal under "IPUs in the Cloud" alongside Gcore Cloud, with quick-start guides for accessing IPUs via Graphcloud.
- Arm HoldingsflagshipBoth Arm and Graphcore are SoftBank AI Computing Segment subsidiaries. Arm's CPU platforms (including the Arm AGI CPU launched in March 2026) and Graphcore's IPU platforms are expected to be linked under SoftBank's AI infrastructure strategy, with Nigel Toon (Graphcore CEO) stating that Graphcore will link up with SoftBank's AI infrastructure.
- SiMa.ai, Arm Holdings, Cerebras Systems, HailominorGrouped with SiMa.ai, Arm Holdings, Cerebras Systems, and Hailo in a peer set of companies profiled for energy-efficient edge AI / Physical AI compute infrastructure.
Scale indicators10 records
Recent moves7 records
Expansion highlights8 records
Graphcore competitors and assessment
Company assessmentBroad incumbents
- Nvidia: Dominant GPU and CUDA platform provider for AI training and inference; the incumbent Graphcore explicitly positions against, with overlapping target customers (enterprises, hyperscalers, cloud aggregators) but vastly broader scale and software moat.
Emerging players
- Blaize: AI compute platform (GSP architecture) for edge and automotive inference; an emerging AI accelerator player competing in similar low-latency, energy-efficient workloads to Graphcore's IPU edge positioning.
- Positron: Memory-focused AI inference accelerator targeting transformer and LLM workloads; an emerging alternative accelerator with comparable design goals to Graphcore's in-processor-memory IPU approach for inference economics.
- SiMa.ai: Edge AI / Physical AI inference SoC company; explicitly grouped with Graphcore in industry coverage of energy-efficient Physical AI infrastructure and competes for the same low-power edge inference design wins.
- Hailo: Edge AI accelerator (Hailo-8, Hailo-10) targeting low-power computer vision and Physical AI inference; listed alongside Graphcore in the same energy-efficient Physical AI infrastructure peer cohort, with overlapping edge and industrial customers.
Direct peers
- SambaNova Systems: Builds the RDU (Reconfigurable Dataflow Unit) DataScale systems as a non-GPU alternative for enterprise AI training and inference; a directly comparable specialty AI accelerator vendor targeting similar on-prem and cloud customers.
- Groq: Pure LPU inference accelerator focused on low-latency, high-throughput AI inference; competes with Graphcore's IPU for the same enterprise inference workloads and is sold into overlapping cloud and AI-as-a-service channels.
- Tenstorrent: AI processor and RISC-V based compute company (backed by Jim Keller) building accelerators for training and inference; directly comparable specialty AI silicon vendor pursuing a similar non-Nvidia positioning.
- Cerebras Systems: Developer of the Wafer-Scale Engine AI accelerator and CS systems targeting large-model training and inference; competes head-to-head with Graphcore IPUs in enterprise and research AI compute, and is also listed in the same energy-efficient Physical AI infrastructure peer cohort.
Others
- Ampere Computing: Arm-based cloud-native CPU company and fellow SoftBank AI Computing Segment subsidiary; not a direct accelerator peer but a directly adjacent SoftBank portfolio company whose integration with Graphcore is explicitly part of SoftBank's AI infrastructure strategy.
Market position
Strengths5 records
Weaknesses5 records
Competitive moat5 records
Key risks6 records
Key highlights7 records
Customer concentration
Graphcore social profiles
Digital presenceGraphcore compliance and trust
Trust signalCompliance4 records
Graphcore financial estimates
Financial estimateRevenue estimate
Valuation estimate
Graphcore leadership team
Management profileNumber of profiles
Profiles12 records
Graphcore funding detail
Funding detailFunding overview
Funding rounds8 records
Investors26 records
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Graphcore M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Graphcore
What does Graphcore do?
Graphcore designs and manufactures Intelligence Processing Units (IPUs) — purpose-built AI processors with an in-processor-memory architecture — and sells them as IPU-M2000 building blocks, Bow Pod and IPU-POD scale-out systems, and C600 PCIe accelerator cards. The hardware is paired with the Poplar SDK (PopART, PopTorch, PopLibs, PopRun, PopDist) and Graphcloud, forming an integrated platform positioned as an alternative to Nvidia GPU/CUDA stacks for AI training and inference workloads. Customers include enterprises, cloud providers, AI researchers, and edge AI builders.
Is Graphcore a public or private company?
Graphcore is a private company. It is classified as corporate owned and is currently acquired.
When was Graphcore founded?
Graphcore was founded in 2016. It employs 501 to 1,000 people.
Where is Graphcore based?
Graphcore is headquartered in Bristol, United Kingdom, in the Europe region.
How does Graphcore make money?
Five revenue lines are on record. Hardware sales — IPU systems are the primary driver. The others are software licensing — Poplar SDK and developer tools, IP licensing / royalties, cloud / managed service offering via partners and professional services — support, training, evaluation.
Who are Graphcore's main competitors?
Nvidia is listed as a broad incumbent. Emerging players are Blaize, Positron, SiMa.ai and Hailo. Direct peers are SambaNova Systems, Groq, Tenstorrent and Cerebras Systems. Ampere Computing is listed as an others.
Does Graphcore have an API?
Yes. V-IPU API (vipu-protos) for Virtual-IPU integration, version 1.18.0. Supports developer access to Graphcore's compute virtualization layer. Software downloads and developer resources are hosted at the Graphcore Developer portal with documentation available at docs.graphcore.ai, and source code samples available via GitHub at github.com/graphcore. Developer documentation is at docs.graphcore.ai.
What industry is Graphcore in?
Graphcore's product category is AI accelerator hardware. Its primary akta.pro industry code is HDAHAJAF, AI/Neural Processing Units (NPUs) & ML Accelerators, with a secondary code of HDAHABAC, Compute & Acceleration ICs (CPU/GPU/AI/FPGA/ASIC). Its NAICS code is 334111 and its SIC code is 3571.