Developer docs
API playgroundTry for free, no card

Search company profiles

Iluvatar CoreX

Full company profile

uuid0000rvw

Namestring
Iluvatar CoreX
Legal namestring
上海天数智芯半导体股份有限公司
Websiteurl
iluvatar.com
Company typeenum
Public
Founded yearint
2018
Descriptiontext

Iluvatar CoreX (天数智芯) is a Chinese fabless semiconductor company that designs and sells general-purpose GPU (GPGPU) chips and full-stack AI computing solutions. Founded in 2015 and headquartered in Shanghai, the company is publicly listed on the Hong Kong Stock Exchange (9903.HK) following its January 2026 IPO. Its product portfolio spans three lines: the TianGai (天垓) training series for cloud AI training workloads (TianGai-100 launched 2021, TianGai-150 with 64GB HBM, and Gen3 entering mass production in Q1 2026); the Zhikai (智铠) inference series for AI inference (Zhikai-50 and Zhikai-100); and the Tongyang (彤央) edge computing series (TY1000, TY1100, TY1100_NX, TY1200) for embedded and embodied AI applications. The company's technology is unified by a proprietary software stack (天数智算软件栈) providing CUDA ecosystem compatibility and a self-developed compiler, operator library, and inference engine, supported by the DeepSpark open-source developer community.

Iluvatar CoreX generates revenue primarily through one-time hardware sales of accelerator cards and modules (89% of 2025 revenue, branded as the 'core GPU business'), supplemented by AI computing solutions and services (customization, model adaptation, deployment support) and a usage-based AI cloud platform. The company sells exclusively to enterprise customers in China — including cloud service providers, financial institutions, hospitals, manufacturers, smart city projects, and robotics firms — via a direct enterprise sales team and a 15+ certified server OEM partner channel. Named flagship accounts include ByteDance (in negotiation for 50,000+ chips), Quzhou People's Hospital, Luckin Coffee, GrLanRuo Robotics, and over 20 pilot cities for vehicle-road-cloud integration programs.

The business has scaled rapidly, with 2025 revenue of RMB 1.03 billion (~$149M), up 92% year-over-year, and core GPU revenue up 149.6%. The company has cumulatively shipped 52,000+ units to 290+ customers across 1,000+ deployments, and reports 56.6% gross margins (2024). It remains unprofitable, with annual losses of approximately RMB 1 billion reflecting heavy R&D investment (R&D headcount is 70%+ of employees). The competitive positioning rests on a regulatory tailwind from US export controls restricting Nvidia, full technology-stack independence from foreign IP, and a four-generation architecture roadmap targeting NVIDIA Rubin performance parity by 2027.

Short descriptiontext

Iluvatar CoreX is a Shanghai-based Chinese fabless semiconductor company that designs self-developed general-purpose GPU chips and AI computing solutions for cloud training, inference, and edge applications, serving enterprise customers including cloud providers, ByteDance, hospitals, and smart city projects.

Operating statusenum
Ipo
Ownership categoryenum
Headcount rangeband
101–250
akta.pro rankint
HeadquartersShanghai, China
HQ citystring
Shanghai
HQ countrystring
China
HQ regionstring
Asia
Markets served

Serves global market

Offices6 records

Each record includes

City, Country, Type, Description, Source

Keyword5 values
general-purpose GPU chips, AI training accelerators, AI inference accelerators, edge computing modules, GPU computing software
Industry6 codes
PRIMARYHDAHABAC
Compute & Acceleration ICs (CPU/GPU/AI/FPGA/ASIC)
SECONDARYHDAAAAAA
AI Accelerators (GPUs/TPUs/NPUs/ASICs)
Hardware, Technology & Telecom/Artificial Intelligence & Machine Learning/AI Infrastructure & Compute (GPUs/ASICs, accelerators)
SECONDARYHDACAGAA
Graphics & Accelerator Cards (GPUs, compute accelerators)
Hardware, Technology & Telecom/Computing Hardware & Peripherals/PC Components & Upgrades (Motherboards, GPUs, RAM, PSUs, Cooling)
SECONDARYHDAAAAAC
AI Inference Hardware & Edge Accelerators
Hardware, Technology & Telecom/Artificial Intelligence & Machine Learning/AI Infrastructure & Compute (GPUs/ASICs, accelerators)
SECONDARYHDAAAJAC
Edge AI Hardware & NPUs (SoCs, modules, accelerators)
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)
NAICS code4 codes
  • Computer and Peripheral Equipment Manufacturing3341
  • Computer Systems Design and Related Services54151
  • Computer Systems Design Services541512
  • Custom Computer Programming Services541511
SIC code3 codes
  • Electronic Computers3571
  • Services-Computer Integrated Systems Design7373
  • Services-Computer Programming Services7371
Product category
AI Semiconductor (GPGPU chips)
No data
GTM motion2 records

Each record includes

Type, Description, Source

Revenue model3 records
1GPU Hardware Product Sales
TypeOne Time License
Description

Sale of proprietary general-purpose GPU chips and accelerator cards (TianGai training series, ZhiKai inference series, Tongyang edge series) to enterprise customers, cloud service providers, data centers, and government/research institutions. One-time hardware purchase model with enterprise procurement cycles. Core GPU business accounted for 89% of total 2025 revenue.

iluvatar.com
2AI Computing Solutions & Services
TypeProfessional Services
Description

Full-stack AI computing solutions including software stack licensing, customization services, technical support, and system integration. The company provides 'ready-to-use' services combining hardware and proprietary software. Professional services include model adaptation, algorithm optimization, and deployment support. As of June 2025, over 5.2万 units shipped to 290+ customers across financial, medical, internet, education, and autonomous driving sectors.

iluvatar.com
3AI Computing Power Services
TypeUsage Based
Description

AI cloud platform providing public cloud access to TianGai-100 GPU for AI painting and other applications. Revenue from compute time consumption on the platform.

electronicsmedia.info
Marketing channels6 records

Each record includes

Title, Type, Stage, Description, Source

Distribution channels4 records

Each record includes

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

Cost components5 values
Technology or R&D, Personnel, Supply Chain, Marketing or Sales, Operations
Pricing details1 tier
1Enterprise GPU hardware and solutions sales via direct quotation
ModelOtherBilling cadenceMulti-year contract
Notes

No public pricing disclosed. All transactions are enterprise B2B quotations through sales team.

iluvatar.com
GTM typeB2B
B2B
Offering typeHardware or Manufacturing
Hardware or Manufacturing
Brand1 of 5 records shown
1天垓 (TianGai)
Description

Training series GPU products for AI training workloads

iluvatar.com
+4 more records
Core offering1 text field

Iluvatar CoreX designs, manufactures, and sells self-developed general-purpose GPU (GPGPU) chips and accelerator cards for AI training, AI inference, and edge computing workloads. The company offers a full-stack portfolio including the TianGai training series (cloud data center GPUs), Zhikai inference series (AI inference accelerators), TongYang edge computing modules (embedded AI for robotics and edge devices), and a proprietary Iluvatar Computing Software Stack that is CUDA-compatible. Products are sold to enterprise customers including cloud service providers, financial institutions, healthcare providers, internet companies, and government/research institutions across China.

Differentiator
Functional benefit
Problem solved
Quantifiable outcome1 of 9 values shown
  • 238% year-to-date stock price appreciation by June 2026, reflecting market validation of commercial trajectory
+8 more records
Product overview1 text field

Iluvatar CoreX (天数智芯) is a Chinese general-purpose GPU semiconductor company offering a three-tier product portfolio: (1) TianGai (天垓) training series for cloud AI training workloads, including TianGai 100 (BI-V100) and TianGai 150 with 64GB HBM; (2) Zhikai (智铠) inference series for AI inference, including Zhikai 50 and Zhikai 100 (MR-V100) acceleration cards supporting up to 128-channel video decoding; and (3) TongYang (彤央) edge computing series for edge AI comprising TY1000, TY1100, TY1100_NX, and TY1200 (300TOPs) modules targeting AIPC, embodied AI, and robotics. The product line is unified by the proprietary Iluvatar Computing Software Stack (天数智算软件栈) providing heterogeneous computing platform with CUDA ecosystem compatibility, and supported by the DeepSpark open-source community for developer ecosystem engagement. The company positions as China's first to achieve mass production of general-purpose GPUs supporting both AI training and inference, serving 290+ customers with 52,000+ cumulative shipments as of mid-2025.

Product and service9 records
1TianGai 100 (天垓100) Cloud Training GPU
CategoryAI training accelerator (cloud GPU)
2TianGai 150 (天垓150) Training GPU
CategoryAI training accelerator (cloud GPU)
3Zhikai 100 (智铠100) Inference GPU
CategoryAI inference accelerator
4TongYang TY1000 (彤央TY1000) Edge Computing Module
CategoryEdge AI computing module
5TongYang TY1100 (彤央TY1100) Edge Computing Module
CategoryEdge AI computing module
6TongYang TY1100_NX (彤央TY1100_NX) Edge Computing Terminal
CategoryEdge AI computing module
7TongYang TY1200 (彤央TY1200) High-Performance Edge AI Terminal
CategoryEdge AI computing module
8Iluvatar Computing Software Stack (天数智算软件栈)
9AI Cloud Platform (gpu.ai-galaxy.cn)
CategoryAI cloud computing service
Scale indicator11 records

Each record includes

Type, Value, Description, Source

Partnership16 partners
1DeepSpark Open Source Community Ecosystem Partners
Strategic tierCoreTypeOthersAnnounced on2026-01-26
Description

DeepSpark is Iluvatar CoreX's proprietary open-source community for AI and general-purpose computing application development. Features the 'Hundred Applications Open Platform' with open-source algorithms and models across AI domains. Community attracts developers and builds ecosystem loyalty around Iluvatar hardware.

iluvatar.com
2Shanghai AI Laboratory (OpenDataLab & DeepLink Teams)
Strategic tierCoreTypeStrategic or Co-development Partner
Description

Iluvatar CoreX deeply contributed to the SuperNode Technology Whitepaper led by Shanghai AI Lab's DeepLink team, jointly with 8 top universities and 16 industry partners. Also completed deep adaptation of MinerU intelligent document parsing engine on Iluvatar computing platform. Strategic collaboration for AI ecosystem standards.

iluvatar.com
3Tsinghua University (Electronic Engineering Department)
Strategic tierCoreTypeStrategic or Co-development Partner
Description

Tsinghua University Electronic Engineering Department conducts deep school-enterprise exchanges with Iluvatar CoreX, including talent cultivation, joint research, and achievement translation. Executive Director Sun Yile (Tsinghua 1997 alumnus) facilitates cooperation. Topics include chip architecture, engineering implementation, and industry applications.

iluvatar.com
4Zhejiang University
Strategic tierCoreTypeStrategic or Co-development Partner
Description

Joint laboratories and scholarships established with Zhejiang University for chip talent development. Part of Iluvatar CoreX's university partnership program to build technology talent pipeline.

iluvatar.com
5Zhongjin Yuneng Education Technology Group (中金育能教育科技集团)
Strategic tierMajorTypeStrategic or Co-development Partner
Description

Strategic cooperation signed December 2021 in smart sports, smart training, and intelligent auxiliary diagnosis. Combines AI, IoT, and big data with professional sports training technology. Iluvatar's TianGai/ZhiKai products power AI-driven sports posture assessment and campus safety applications.

iluvatar.com
6Yilu Hui & Hailai Cloud Intelligence (医路汇 & 海莱云智)
Strategic tierMajorTypeImplementation/ SI/ Consulting Partner
Description

Iluvatar CoreX partnered with Yilu Hui and Hailai Cloud Intelligence to build DeepSeek一体机 AI medical solution. ZhiKai-100 GPU provides compute core for Quzhou People's Hospital's local DeepSeek deployment covering electronic medical records, patient analysis, and clinical decision support.

iluvatar.com
Strategic tierFlagshipTypeChannel Partner/ Reseller/ Distributor
Description

ByteDance reportedly purchasing at least 50,000 AI inference chips from Iluvatar CoreX for Doubao chatbot and cloud infrastructure. In-talks deal as of June 2026. Would make Iluvatar ByteDance's third major domestic GPU supplier alongside Huawei and Cambricon. Massive commercial win shifting company from government procurement to commercial enterprise.

8GrLanRuo Robotics (格蓝若机器人)
Strategic tierMajorTypeStrategic or Co-development Partner
Description

GrLanRuo Robotics uses Iluvatar CoreX's Tongyang edge computing products as the 'brain' for embodied AI robots in industrial and production line automation scenarios.

iluvatar.com
Strategic tierMajorTypeChannel Partner/ Reseller/ Distributor
Description

Thousands of Luckin Coffee stores deployed Iluvatar CoreX's Tongyang solutions for video stream processing and consumer data value mining in commercial intelligence applications.

Strategic tierCoreTypeTechnology or Integration
Description

Iluvatar CoreX completed 'flash adaptation' of Zhipu AI's GLM-5 flagship model (744B parameters) immediately upon its release in February 2026. Achieved 'release-day compatibility, adapted means usable' with full-chain deep adaptation of the MoE architecture model.

Strategic tierMajorTypeTechnology or Integration
Description

DeepSeek-R1-70B model deployed locally on Iluvatar CoreX's ZhiKai-100 GPU for AI medical solutions. Supports trillion-parameter model inference with flexible multi-machine cluster expansion capability.

Strategic tierCoreTypeStrategic or Co-development Partner
Description

DeepLink team under Shanghai AI Lab led development of SuperNode Technology Whitepaper with Iluvatar CoreX as one of 8 universities and 16 industry partners. Iluvatar contributed technology expertise for AI supercomputing cluster architecture standards.

Strategic tierMinorTypeOthers
Description

Iluvatar CoreX semiconductor project is among 30 major projects in Chongqing's 100,000 sqm AI innovation hub launched May 2026. Project attracts over 10 leading enterprises and 30 major projects targeting intelligent vehicles, embodied AI, and industry applications.

1420+ Leading Pilot Cities for Vehicle-Road-Cloud Integration
Strategic tierMajorTypeStrategic or Co-development Partner
Description

Iluvatar CoreX collaborates with 20+ leading pilot cities on 'vehicle-road-cloud integration' (车路云一体化) smart transportation initiatives, verifying vehicle-road coordination solutions using Tongyang edge computing products.

chinadaily.com.cn
15New H3C (新华三技术有限公司)
Strategic tierMajorTypeChannel Partner/ Reseller/ Distributor
Description

New H3C certified servers R4900 G6 and R5300 G5 for compatibility with Iluvatar CoreX GPUs. Certified configurations include up to 4-GPU and 8-GPU server deployments.

iluvatar.com
16Server OEM Ecosystem (安擎, 宝德, 长城超云, 百信, etc.)
Strategic tierMajorTypeChannel Partner/ Reseller/ Distributor
Description

Certified server compatibility with 15+ domestic server brands including 安擎 (EG520-G30, EG940H-G22), 宝德, 长城超云, 百信, 天固信安, 苏州超集, 融科联创, 容天汇海, 恒越, 南京坤前, and others. GPUs pre-validated on certified server models.

iluvatar.com
Recent move15 records

Each record includes

Date, Type, Title, Description, Source

Expansion highlight6 records

Each record includes

Type, Description

Peers10 records
TypeDirect peer
Description

Huawei's Ascend series (310/910/910B/910C) is the leading domestic Chinese AI training accelerator, shipping approximately 812,000 AI chips in 2025. Both target the same Chinese AI training/inference workloads with similar domestic substitution positioning.

TypeDirect peer
Description

Cambricon is China's other leading listed AI chip designer (688256.SH) and a fellow ByteDance domestic GPU supplier alongside Huawei and potentially Iluvatar CoreX. Both target Chinese cloud and edge AI inference workloads with proprietary architectures.

3Biren Technology
TypeDirect peer
Description

Biren Technology is a fellow Chinese GPGPU designer that also listed on the Hong Kong Stock Exchange in January 2026 alongside Iluvatar CoreX, both raising combined multi-billion HKD to fund domestic AI chip competition. Both target NVIDIA replacement in Chinese data centers.

TypeDirect peer
Description

Moore Threads is a Chinese fabless GPU designer developing both AI accelerators and graphics cards. Both companies compete for the same domestic Chinese GPU TAM in training, inference, and edge workloads.

TypeDirect peer
Description

MetaX is a Chinese AI GPU startup developing C-series accelerators for training and inference. Both target the same Chinese domestic AI chip market with proprietary GPGPU architectures.

TypeBroad incumbent
Description

NVIDIA is the global incumbent whose H100/H200/B200/Rubin series dominate the AI accelerator market. Iluvatar CoreX explicitly targets NVIDIA architecture parity by 2027 and competes directly with NVIDIA in the Chinese market where export controls have created a regulatory moat.

TypeBroad incumbent
Description

AMD's Instinct MI300/MI325 series competes in the same data center AI accelerator category as Iluvatar's TianGai training series. While AMD's direct China presence is limited, it represents the broader competitive set in the GPGPU category.

TypeEmerging player
Description

Horizon Robotics is a Chinese AI chip designer focused on autonomous driving edge inference, overlapping with Iluvatar CoreX's TongYang edge series in the vehicle-road-cloud integration use case. Both target Chinese automotive and smart transportation edge AI applications.

TypeEmerging player
Description

Hailo develops edge AI processors for computer vision at the edge, overlapping with Iluvatar CoreX's TongYang TY series for embedded AI applications in industrial, retail, and smart city segments.

TypeEmerging player
Description

Black Sesame Technologies is a Chinese AI chip company focused on autonomous driving perception SoCs. It overlaps with Iluvatar CoreX in Chinese smart mobility and edge AI applications, though its focus is narrower (automotive only).

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 risks7 records

Each record includes

Headline, Details, Source

Key highlights7 records

Each record includes

Headline, Details, Source

Customer concentration

Classification, Details

Named customers6 records

Each record includes

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

Segment9 records

Each record includes

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

Ideal customer profile4 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 capability13 records

Each record includes

Type, Description, Source

AI maturity
App detail

Has app

Feature7 records

Each record includes

Title, Differentiator, Description, Source

Core technology
Revenue estimate
Valuation estimate
Number of profiles
Profiles9 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

Investors14 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 →

Iluvatar CoreX

AI Semiconductor (GPGPU chips)iluvatar.com

Iluvatar CoreX is a Shanghai-based Chinese fabless semiconductor company that designs self-developed general-purpose GPU chips and AI computing solutions for cloud training, inference, and edge applications, serving enterprise customers including cloud providers, ByteDance, hospitals, and smart city projects.

What Iluvatar CoreX does

Iluvatar CoreX (天数智芯) is a Chinese fabless semiconductor company that designs and sells general-purpose GPU (GPGPU) chips and full-stack AI computing solutions. Founded in 2015 and headquartered in Shanghai, the company is publicly listed on the Hong Kong Stock Exchange (9903.HK) following its January 2026 IPO. Its product portfolio spans three lines: the TianGai (天垓) training series for cloud AI training workloads (TianGai-100 launched 2021, TianGai-150 with 64GB HBM, and Gen3 entering mass production in Q1 2026); the Zhikai (智铠) inference series for AI inference (Zhikai-50 and Zhikai-100); and the Tongyang (彤央) edge computing series (TY1000, TY1100, TY1100_NX, TY1200) for embedded and embodied AI applications. The company's technology is unified by a proprietary software stack (天数智算软件栈) providing CUDA ecosystem compatibility and a self-developed compiler, operator library, and inference engine, supported by the DeepSpark open-source developer community.

Iluvatar CoreX generates revenue primarily through one-time hardware sales of accelerator cards and modules (89% of 2025 revenue, branded as the 'core GPU business'), supplemented by AI computing solutions and services (customization, model adaptation, deployment support) and a usage-based AI cloud platform. The company sells exclusively to enterprise customers in China — including cloud service providers, financial institutions, hospitals, manufacturers, smart city projects, and robotics firms — via a direct enterprise sales team and a 15+ certified server OEM partner channel. Named flagship accounts include ByteDance (in negotiation for 50,000+ chips), Quzhou People's Hospital, Luckin Coffee, GrLanRuo Robotics, and over 20 pilot cities for vehicle-road-cloud integration programs.

The business has scaled rapidly, with 2025 revenue of RMB 1.03 billion (~$149M), up 92% year-over-year, and core GPU revenue up 149.6%. The company has cumulatively shipped 52,000+ units to 290+ customers across 1,000+ deployments, and reports 56.6% gross margins (2024). It remains unprofitable, with annual losses of approximately RMB 1 billion reflecting heavy R&D investment (R&D headcount is 70%+ of employees). The competitive positioning rests on a regulatory tailwind from US export controls restricting Nvidia, full technology-stack independence from foreign IP, and a four-generation architecture roadmap targeting NVIDIA Rubin performance parity by 2027.

Iluvatar CoreX firmographics

Firmographics
Name
Iluvatar CoreX
Legal name
上海天数智芯半导体股份有限公司
Website
https://iluvatar.com
Company type
Public
Founded year
2018
Operating status
Ipo
Headcount range
101–250 employees
Short description
Iluvatar CoreX is a Shanghai-based Chinese fabless semiconductor company that designs self-developed general-purpose GPU chips and AI computing solutions for cloud training, inference, and edge applications, serving enterprise customers including cloud providers, ByteDance, hospitals, and smart city projects.
Ownership category
akta.pro rank

Iluvatar CoreX industry classification

Industry
Product category
AI Semiconductor (GPGPU chips)
NAICS
Computer and Peripheral Equipment Manufacturing (3341), Computer Systems Design and Related Services (54151), Computer Systems Design Services (541512), Custom Computer Programming Services (541511)
SIC
Electronic Computers (3571), Services-Computer Integrated Systems Design (7373), Services-Computer Programming Services (7371)
akta.pro industry
PRIMARYHDAHABAC
Compute & Acceleration ICs (CPU/GPU/AI/FPGA/ASIC)
SECONDARYHDAAAAAA
AI Accelerators (GPUs/TPUs/NPUs/ASICs)
Hardware, Technology & Telecom/Artificial Intelligence & Machine Learning/AI Infrastructure & Compute (GPUs/ASICs, accelerators)
SECONDARYHDACAGAA
Graphics & Accelerator Cards (GPUs, compute accelerators)
Hardware, Technology & Telecom/Computing Hardware & Peripherals/PC Components & Upgrades (Motherboards, GPUs, RAM, PSUs, Cooling)
SECONDARYHDAAAAAC
AI Inference Hardware & Edge Accelerators
Hardware, Technology & Telecom/Artificial Intelligence & Machine Learning/AI Infrastructure & Compute (GPUs/ASICs, accelerators)
SECONDARYHDAAAJAC
Edge AI Hardware & NPUs (SoCs, modules, accelerators)
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

  • General-purpose GPU chips
  • AI training accelerators
  • AI inference accelerators
  • Edge computing modules
  • GPU computing software

Where Iluvatar CoreX is headquartered

Location

Headquarters

HQ city
Shanghai
HQ country
China
HQ region
Asia

Offices6 records

Markets served

Iluvatar CoreX business model

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

Revenue model

  1. GPU Hardware Product Sales: Sale of proprietary general-purpose GPU chips and accelerator cards (TianGai training series, ZhiKai inference series, Tongyang edge series) to enterprise customers, cloud service providers, data centers, and government/research institutions. One-time hardware purchase model with enterprise procurement cycles. Core GPU business accounted for 89% of total 2025 revenue.
  2. AI Computing Solutions & Services: Full-stack AI computing solutions including software stack licensing, customization services, technical support, and system integration. The company provides 'ready-to-use' services combining hardware and proprietary software. Professional services include model adaptation, algorithm optimization, and deployment support. As of June 2025, over 5.2万 units shipped to 290+ customers across financial, medical, internet, education, and autonomous driving sectors.
  3. AI Computing Power Services: AI cloud platform providing public cloud access to TianGai-100 GPU for AI painting and other applications. Revenue from compute time consumption on the platform.

Pricing tiers

ModelBillingPrice
OtherMulti-year contractEnterprise GPU hardware and solutions sales via direct quotation

Go-to-market motion2 records

Distribution channels4 records

Marketing channels6 records

Iluvatar CoreX product offering

Product offering

Core offering

Iluvatar CoreX designs, manufactures, and sells self-developed general-purpose GPU (GPGPU) chips and accelerator cards for AI training, AI inference, and edge computing workloads. The company offers a full-stack portfolio including the TianGai training series (cloud data center GPUs), Zhikai inference series (AI inference accelerators), TongYang edge computing modules (embedded AI for robotics and edge devices), and a proprietary Iluvatar Computing Software Stack that is CUDA-compatible. Products are sold to enterprise customers including cloud service providers, financial institutions, healthcare providers, internet companies, and government/research institutions across China.

Product overview

Iluvatar CoreX (天数智芯) is a Chinese general-purpose GPU semiconductor company offering a three-tier product portfolio: (1) TianGai (天垓) training series for cloud AI training workloads, including TianGai 100 (BI-V100) and TianGai 150 with 64GB HBM; (2) Zhikai (智铠) inference series for AI inference, including Zhikai 50 and Zhikai 100 (MR-V100) acceleration cards supporting up to 128-channel video decoding; and (3) TongYang (彤央) edge computing series for edge AI comprising TY1000, TY1100, TY1100_NX, and TY1200 (300TOPs) modules targeting AIPC, embodied AI, and robotics. The product line is unified by the proprietary Iluvatar Computing Software Stack (天数智算软件栈) providing heterogeneous computing platform with CUDA ecosystem compatibility, and supported by the DeepSpark open-source community for developer ecosystem engagement. The company positions as China's first to achieve mass production of general-purpose GPUs supporting both AI training and inference, serving 290+ customers with 52,000+ cumulative shipments as of mid-2025.

Differentiator

Problem solved

Functional benefit

Brands

  • 天垓 (TianGai): Training series GPU products for AI training workloads
  • 智铠 (Zhikai)
  • 彤央 (Tongyang)
  • DeepSpark
  • 天数智算软件栈

Products and services

  • TianGai 100 (天垓100) Cloud Training GPU
  • TianGai 150 (天垓150) Training GPU
  • Zhikai 100 (智铠100) Inference GPU
  • TongYang TY1000 (彤央TY1000) Edge Computing Module
  • TongYang TY1100 (彤央TY1100) Edge Computing Module
  • TongYang TY1100_NX (彤央TY1100_NX) Edge Computing Terminal
  • TongYang TY1200 (彤央TY1200) High-Performance Edge AI Terminal
  • Iluvatar Computing Software Stack (天数智算软件栈)
  • AI Cloud Platform (gpu.ai-galaxy.cn)

Quantifiable outcome

  • 238% year-to-date stock price appreciation by June 2026, reflecting market validation of commercial trajectory
  • +8 more outcomes

Companies that use Iluvatar CoreX

Customer profile

Named customers6 records

Segments9 records

Ideal customer profiles4 records

Iluvatar CoreX technology and API

Technology

Technology focussed Yes

API detail

Has API
No
API docs
API detail

Core technology

AI maturity

App detail

AI capability13 records

Feature7 records

Iluvatar CoreX partnerships and signals

Strategic signal

Partnerships

16 partnerships are on record, tiered core, major, flagship and minor.

  • DeepSpark Open Source Community Ecosystem PartnerscoreOthers · 26 January 2026DeepSpark is Iluvatar CoreX's proprietary open-source community for AI and general-purpose computing application development. Features the 'Hundred Applications Open Platform' with open-source algorithms and models across AI domains. Community attracts developers and builds ecosystem loyalty around Iluvatar hardware.
  • Shanghai AI Laboratory (OpenDataLab & DeepLink Teams)coreStrategic or Co-development PartnerIluvatar CoreX deeply contributed to the SuperNode Technology Whitepaper led by Shanghai AI Lab's DeepLink team, jointly with 8 top universities and 16 industry partners. Also completed deep adaptation of MinerU intelligent document parsing engine on Iluvatar computing platform. Strategic collaboration for AI ecosystem standards.
  • Tsinghua University (Electronic Engineering Department)coreStrategic or Co-development PartnerTsinghua University Electronic Engineering Department conducts deep school-enterprise exchanges with Iluvatar CoreX, including talent cultivation, joint research, and achievement translation. Executive Director Sun Yile (Tsinghua 1997 alumnus) facilitates cooperation. Topics include chip architecture, engineering implementation, and industry applications.
  • Zhejiang UniversitycoreStrategic or Co-development PartnerJoint laboratories and scholarships established with Zhejiang University for chip talent development. Part of Iluvatar CoreX's university partnership program to build technology talent pipeline.
  • Zhongjin Yuneng Education Technology Group (中金育能教育科技集团)majorStrategic or Co-development PartnerStrategic cooperation signed December 2021 in smart sports, smart training, and intelligent auxiliary diagnosis. Combines AI, IoT, and big data with professional sports training technology. Iluvatar's TianGai/ZhiKai products power AI-driven sports posture assessment and campus safety applications.
  • Yilu Hui & Hailai Cloud Intelligence (医路汇 & 海莱云智)majorImplementation/ SI/ Consulting PartnerIluvatar CoreX partnered with Yilu Hui and Hailai Cloud Intelligence to build DeepSeek一体机 AI medical solution. ZhiKai-100 GPU provides compute core for Quzhou People's Hospital's local DeepSeek deployment covering electronic medical records, patient analysis, and clinical decision support.
  • ByteDanceflagshipChannel Partner/ Reseller/ DistributorByteDance reportedly purchasing at least 50,000 AI inference chips from Iluvatar CoreX for Doubao chatbot and cloud infrastructure. In-talks deal as of June 2026. Would make Iluvatar ByteDance's third major domestic GPU supplier alongside Huawei and Cambricon. Massive commercial win shifting company from government procurement to commercial enterprise.
  • GrLanRuo Robotics (格蓝若机器人)majorStrategic or Co-development PartnerGrLanRuo Robotics uses Iluvatar CoreX's Tongyang edge computing products as the 'brain' for embodied AI robots in industrial and production line automation scenarios.
  • Luckin Coffee (瑞幸咖啡)majorChannel Partner/ Reseller/ DistributorThousands of Luckin Coffee stores deployed Iluvatar CoreX's Tongyang solutions for video stream processing and consumer data value mining in commercial intelligence applications.
  • Zhipu AI (智谱AI)coreTechnology or IntegrationIluvatar CoreX completed 'flash adaptation' of Zhipu AI's GLM-5 flagship model (744B parameters) immediately upon its release in February 2026. Achieved 'release-day compatibility, adapted means usable' with full-chain deep adaptation of the MoE architecture model.
  • DeepSeekmajorTechnology or IntegrationDeepSeek-R1-70B model deployed locally on Iluvatar CoreX's ZhiKai-100 GPU for AI medical solutions. Supports trillion-parameter model inference with flexible multi-machine cluster expansion capability.
  • DeepLink (上海人工智能实验室)coreStrategic or Co-development PartnerDeepLink team under Shanghai AI Lab led development of SuperNode Technology Whitepaper with Iluvatar CoreX as one of 8 universities and 16 industry partners. Iluvatar contributed technology expertise for AI supercomputing cluster architecture standards.
  • Chongqing Liangjiang New AreaminorOthersIluvatar CoreX semiconductor project is among 30 major projects in Chongqing's 100,000 sqm AI innovation hub launched May 2026. Project attracts over 10 leading enterprises and 30 major projects targeting intelligent vehicles, embodied AI, and industry applications.
  • 20+ Leading Pilot Cities for Vehicle-Road-Cloud IntegrationmajorStrategic or Co-development PartnerIluvatar CoreX collaborates with 20+ leading pilot cities on 'vehicle-road-cloud integration' (车路云一体化) smart transportation initiatives, verifying vehicle-road coordination solutions using Tongyang edge computing products.
  • New H3C (新华三技术有限公司)majorChannel Partner/ Reseller/ DistributorNew H3C certified servers R4900 G6 and R5300 G5 for compatibility with Iluvatar CoreX GPUs. Certified configurations include up to 4-GPU and 8-GPU server deployments.
  • Server OEM Ecosystem (安擎, 宝德, 长城超云, 百信, etc.)majorChannel Partner/ Reseller/ DistributorCertified server compatibility with 15+ domestic server brands including 安擎 (EG520-G30, EG940H-G22), 宝德, 长城超云, 百信, 天固信安, 苏州超集, 融科联创, 容天汇海, 恒越, 南京坤前, and others. GPUs pre-validated on certified server models.

Scale indicators11 records

Recent moves15 records

Expansion highlights6 records

Iluvatar CoreX competitors and assessment

Company assessment

Direct peers

  • Huawei (Ascend): Huawei's Ascend series (310/910/910B/910C) is the leading domestic Chinese AI training accelerator, shipping approximately 812,000 AI chips in 2025. Both target the same Chinese AI training/inference workloads with similar domestic substitution positioning.
  • Cambricon Technologies: Cambricon is China's other leading listed AI chip designer (688256.SH) and a fellow ByteDance domestic GPU supplier alongside Huawei and potentially Iluvatar CoreX. Both target Chinese cloud and edge AI inference workloads with proprietary architectures.
  • Biren Technology: Biren Technology is a fellow Chinese GPGPU designer that also listed on the Hong Kong Stock Exchange in January 2026 alongside Iluvatar CoreX, both raising combined multi-billion HKD to fund domestic AI chip competition. Both target NVIDIA replacement in Chinese data centers.
  • Moore Threads: Moore Threads is a Chinese fabless GPU designer developing both AI accelerators and graphics cards. Both companies compete for the same domestic Chinese GPU TAM in training, inference, and edge workloads.
  • MetaX (沐曦集成电路): MetaX is a Chinese AI GPU startup developing C-series accelerators for training and inference. Both target the same Chinese domestic AI chip market with proprietary GPGPU architectures.

Broad incumbents

  • NVIDIA: NVIDIA is the global incumbent whose H100/H200/B200/Rubin series dominate the AI accelerator market. Iluvatar CoreX explicitly targets NVIDIA architecture parity by 2027 and competes directly with NVIDIA in the Chinese market where export controls have created a regulatory moat.
  • AMD: AMD's Instinct MI300/MI325 series competes in the same data center AI accelerator category as Iluvatar's TianGai training series. While AMD's direct China presence is limited, it represents the broader competitive set in the GPGPU category.

Emerging players

  • Horizon Robotics: Horizon Robotics is a Chinese AI chip designer focused on autonomous driving edge inference, overlapping with Iluvatar CoreX's TongYang edge series in the vehicle-road-cloud integration use case. Both target Chinese automotive and smart transportation edge AI applications.
  • Hailo: Hailo develops edge AI processors for computer vision at the edge, overlapping with Iluvatar CoreX's TongYang TY series for embedded AI applications in industrial, retail, and smart city segments.
  • Black Sesame Technologies: Black Sesame Technologies is a Chinese AI chip company focused on autonomous driving perception SoCs. It overlaps with Iluvatar CoreX in Chinese smart mobility and edge AI applications, though its focus is narrower (automotive only).

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks7 records

Key highlights7 records

Customer concentration

Iluvatar CoreX financial estimates

Financial estimate

Revenue estimate

Valuation estimate

Iluvatar CoreX leadership team

Management profile

Number of profiles

Profiles9 records

Iluvatar CoreX funding detail

Funding detail

Funding overview

Funding rounds5 records

Investors14 records

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

Iluvatar CoreX 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 Iluvatar CoreX

What does Iluvatar CoreX do?

Iluvatar CoreX designs, manufactures, and sells self-developed general-purpose GPU (GPGPU) chips and accelerator cards for AI training, AI inference, and edge computing workloads. The company offers a full-stack portfolio including the TianGai training series (cloud data center GPUs), Zhikai inference series (AI inference accelerators), TongYang edge computing modules (embedded AI for robotics and edge devices), and a proprietary Iluvatar Computing Software Stack that is CUDA-compatible. Products are sold to enterprise customers including cloud service providers, financial institutions, healthcare providers, internet companies, and government/research institutions across China.

Is Iluvatar CoreX a public or private company?

Iluvatar CoreX is a public company. It is classified as public and is currently ipo.

When was Iluvatar CoreX founded?

Iluvatar CoreX was founded in 2018. It employs 101 to 250 people.

Where is Iluvatar CoreX based?

Iluvatar CoreX is headquartered in Shanghai, China, in the Asia region.

How does Iluvatar CoreX make money?

Three revenue lines are on record. GPU Hardware Product Sales are the primary driver. The others are AI Computing Solutions & Services and AI Computing Power Services.

Who are Iluvatar CoreX's main competitors?

Direct peers on record are Huawei (Ascend), Cambricon Technologies, Biren Technology, Moore Threads and MetaX (沐曦集成电路). Broad incumbents are NVIDIA and AMD. Emerging players are Horizon Robotics, Hailo and Black Sesame Technologies.

Does Iluvatar CoreX have an API?

No public API is recorded for Iluvatar CoreX.

What industry is Iluvatar CoreX in?

Iluvatar CoreX's product category is AI Semiconductor (GPGPU chips). Its primary akta.pro industry code is HDAHABAC, Compute & Acceleration ICs (CPU/GPU/AI/FPGA/ASIC), with a secondary code of HDAAAAAA, AI Accelerators (GPUs/TPUs/NPUs/ASICs). Its NAICS code is 3341 and its SIC code is 3571.

Unlock the full company data

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

Contact sales
Live signals
South China Morning PostChina’s AI chipmakers report first profits, but will the trend continue?China's pure-play AI chipmakers turned profitable in the third quarter, with domestic GPU developers like Moore Threads, MetaX, and Iluvatar CoreX filling the void left by Nvidia. However, soaring memory costs and supply-chain difficulties are expected to limit long-term growth, prompting investment banks to cut margin forecasts.WccftechChina’s Iluvatar CoreX Claims Its Tiangai 300 GPU Beats NVIDIA’s H200 In Its Own Live Demo, But The Only Outside Test Shows Just A 5% LeadIluvatar CoreX demonstrated its Tiangai 300 GPU beating NVIDIA's H200 in text-to-image generation, completing tasks at 0.101 images per second versus H200's 0.086. Third-party testing showed a 5% throughput lead, and the chip delivered 1.33 times H100 training throughput. The firm aims to leapfrog NVIDIA's Rubin platform within two years.South China Morning PostChina’s Iluvatar CoreX says new GPU matches Nvidia’s H200 as chip gap closesIluvatar CoreX claimed its Tiangai 300 GPU outperformed Nvidia's H200 in complex multimodal tasks, including text-to-image and text-to-video generation using frontier AI models. The demonstration was shown to mainland Chinese media outlets on Wednesday.South China Morning PostIluvatar CoreX’s first AI computing supernode squeezes most GPUs in 1 systemIluvatar CoreX unveiled its first supernode, the Tiangai 300 IXL144, at Shanghai's CIIF, packing 144 GPUs and 18 CPUs into one cabinet. The system aims to lower token costs by 40% versus mainstream alternatives, offering a local alternative amid US export controls.LeiphoneExclusive interpretation丨Four semi-annual reports of domestic GPUs have been released, and only after making money did they realize that the secondary market is "hard to please".Four domestic GPU companies released semi-annual reports, showing mixed profitability: Biran Technology's revenue grew nearly 20 times but it remains in the red, while Tian Shu Zhi Xin and Mu Xi turned losses into profits, and Mo Er Thread is near breakeven. Analysts say the market now prioritizes sustainable revenue growth and order quality over one-time profits.Investing.comChina AI chipmakers hike prices as HBM shortage hits domestic push - Reuters By Investing.comChinese AI chipmakers Huawei and Cambricon raised prices for processors as a global high-bandwidth memory shortage squeezes domestic alternatives to Nvidia. Huawei's Ascend 950DT price rose 20-50% to over 250,000 yuan, while Cambricon's 690 chip increased 20-30%. The shortage forces suppliers to prioritize customers, including doubling Iluvatar CoreX's GPU shipments to ByteDance.Investing.comExclusive-China’s AI chipmakers raise prices as high-bandwidth memory shortage bites By ReutersChinese AI chipmakers Huawei, Cambricon, MetaX, and Iluvatar CoreX raised prices for AI processors by 20-50% due to high-bandwidth memory shortages. Huawei's Ascend 950DT card now costs over 250,000 yuan, up from earlier levels. The hikes reflect reliance on expensive grey-market HBM after U.S. export controls.Qatar TribuneChina’s AI chipmakers pay more for a crucial ingredientChinese AI chipmakers Huawei and Cambricon raised prices for their processors, with Huawei's Ascend 950DT card now above 250,000 yuan, a 20-50% increase. The hikes reflect soaring high-bandwidth memory costs from export controls, forcing reliance on grey-market supplies. The shortage is reshaping allocations, with Iluvatar CoreX doubling GPU shipments to ByteDance.Caixin GlobalChart of the Day: Rising Revenues Mask Profit Hurdles for China’s GPU QuartetChina's four leading GPU developers—MetaX, Iluvatar CoreX, Biren, and Moore Threads—crossed or approached 1 billion yuan revenue in H1 2026, driven by domestic AI demand. Moore Threads led with 1.7 billion yuan, while Biren saw nearly 2,000% growth, but profitability remains uneven, with MetaX and Iluvatar turning profitable and Biren and Moore Threads still in the red.South China Morning PostChina’s AI chipmakers post surging sales in tech self-reliance driveChinese AI chipmakers posted mixed first-half results as investors reacted to earnings reports, with MetaX and Iluvatar moving to profit while Biren narrowed losses. MetaX, which debuted on Shanghai's Star Market in December, posted a net profit of 612 million yuan, reversing a 186 million yuan loss, with revenue up 44.7 per cent to 1.32 billion yuan.