Rising AI demand and localization trends support Iluvatar's growth outlook

Overview

This report is based on Goldman Sachs' recent discussion with Iluvatar's chairman in Shanghai and focuses on the company's growth outlook in China's GPU and AI computing solutions market. Management remained positive on the company's future growth, mainly citing the localization trend amid geopolitical tensions, the ramp-up in shipments of GPGPU solutions, upgrades to AI training and inference products, and expanding demand across multiple industries.

Core views

The key view is that the continued launch of Chinese domestic AI foundation models and applications will drive higher AI compute spending and GPU demand. As a domestic GPU supplier, Iluvatar already has full-stack capabilities spanning architecture design, hardware integration, software optimization, and post-deployment support, and serves AI training through its TG series and AI inference through its ZK series. The company's customer base includes CSPs, AI model developers, and enterprise clients across sectors such as financial services, healthcare, and transportation.

Analysis framework

The report uses a management visit and industry read-through approach to map Iluvatar's operating commentary onto demand trends in China's GPU industry. The analysis emphasizes product generation upgrades, customer base expansion, coverage of AI training and inference use cases, and demand elasticity from domestic substitution and local AI ecosystem development.

Methodology notes

• Company researchChairman visit

  Management discussion

  By speaking with the company's chairman, the team obtained management's view on demand, product roadmap, and customer expansion.

• Industry mappingRead-across

  Chinese GPU industry demand mapping

  Map Iluvatar management's statements on AI demand, localization, and product upgrades to the growth trend in China's GPU industry.

• Goldman Sachs disclosure frameworkGS Factor Profile

  Growth, financial returns, valuation multiples, and overall percentile comparison

  The disclosure section states that Goldman Sachs compares stocks using growth, financial returns, valuation multiples, and composite indicators, but this report does not assign Iluvatar a coverage rating or target price.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Iluvatar (9903.HK)

  Core company in the report, Not Covered

  Strengths

  Leading domestic GPU supplier; offers GPGPU accelerators, servers, clusters, and full-stack AI computing solutions; more than 290 customers; supports 900+ AI deployments and applications.

  Weaknesses

  The report provides no financial data, earnings forecasts, valuation, target price, or rating, leaving the investment case without quantitative support.

  Comparison

  Compared with a typical AI hardware supplier, the company emphasizes both training and inference product lines, as well as full-stack capabilities from hardware to software optimization and post-deployment support.

  Risks

  AI demand may fall short of expectations, product iteration may lag, customer adoption may be slower than expected, and geopolitical and supply-chain constraints may persist.

• China GPU industry

  Industry read-through target

  Strengths

  The continued launch of local AI foundation models and applications, localization trends, and the build-out of large-scale training clusters support demand.

  Weaknesses

  The industry may still be affected by technology gaps, ecosystem maturity, supply chains, and customer procurement pace.

  Comparison

  The report views Iluvatar management's positive feedback as evidence of rising demand in China's GPU market.

  Risks

  AI capex may be volatile, domestic substitution may progress more slowly than expected, and overseas restrictions or intensifying competition may weigh on the outlook.

Key data

• Ticker9903.HKIluvatar is marked as Not Covered in the report.
• Customer countmore than 290As of 2Q25, customers include CSPs, AI model developers, and enterprise clients.
• AI deployments and application support900+Covers financial services, healthcare, transportation, and other scenarios.
• AI training productsTG Gen 1 to TG Gen 3The company is developing the next-generation TG GPU product, targeting large-scale AI training demand.
• AI inference productsZK seriesEmphasizes edge and cloud inference performance, easy deployment, and video processing capabilities.

Impact & implications

If China's local AI ecosystem continues to expand and AI foundation models and applications drive build-out of compute clusters, Iluvatar could benefit from domestic GPU procurement, AI training demand, and the expansion of inference deployment scenarios. The report is positive on demand in China's GPU industry on a read-through basis, but since Iluvatar is Not Covered, no formal investment rating or target price is provided.

Risks

• The report does not provide Iluvatar financial forecasts, valuation, or target price, limiting the quantifiability of the investment conclusion.
• Growth in AI training and inference demand may fall short of management expectations.
• The development and mass production schedule for the next-generation TG GPU product may face uncertainty.
• The pace of domestic GPU ecosystem development, software adaptation, and customer deployment may affect commercialization momentum.
• Geopolitical, supply-chain, and regulatory changes may affect product delivery and customer procurement.

What to watch

• Performance, launch timing, and customer adoption progress of the next-generation TG GPU product.
• Deployment speed of the ZK series in edge and cloud AI inference scenarios.
• Whether customer counts continue to rise and the order quality from CSPs, AI model developers, and enterprise clients.
• Whether the 900+ AI deployments and applications can translate into sustainable revenue and profit.
• The capex trend in China's local AI foundation models, applications, and compute cluster build-out.