HSBC maintains Hold rating on Z.AI and lowers target price to HKD1,500

Overview

This is an HSBC company research report on Z.AI (2513.HK), focused on reassessing competition among China's frontier AI models, Z.AI's ARR growth, computing supply, and valuation following the launch of Kimi K3. The report maintains a Hold rating while lowering the target price from HKD1,900 to HKD1,500.

Core views

HSBC's core view is that Z.AI's commercialization is progressing better than expected, with ARR reaching USD1bn ahead of schedule and potentially reaching USD2bn by December 2026 and USD6.5bn by December 2027. The company may also achieve profitability in 2027, one year earlier than previously expected. However, competition is intensifying significantly. Kimi K3 shows that models with larger parameter scales will drive rapid capability iteration, making the lead of frontier models more transient. Pricing pressure, a low-margin API revenue mix, share dilution, and a declining scarcity premium together weigh on valuation.

Analysis framework

The report combines fundamental forecasts, peer ARR growth comparisons, OpenRouter token usage trends, model parameter scale and capability iteration, computing supply constraints, and DCF valuation to assess the balance between Z.AI's improving growth and competitive pressure.

Methodology notes

• Valuation methodsDCF

  10-year discounted cash flow model

  HSBC uses a 10-year DCF model to derive a target price of HKD1,500, with key assumptions including an 11.8% WACC and a 3% terminal growth rate.

• Business metricsARR

  Annual recurring revenue

  The report uses the revised ARR trajectory to measure the commercialization progress of Z.AI's MaaS and API businesses and compares its growth and valuation multiples with frontier AI companies such as Anthropic.

• Technology assessmentScaling law

  Relationship between model parameter scale and capability improvement

  The report believes that Kimi K3's intelligence leap over K2.7 mainly came from increasing the parameter scale from 1trn to 2.8trn, indicating that larger-parameter models may still drive capability improvements.

Asset mapping & comparison

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

• Z.AI 2513.HK

  Covered company

  Strengths

  ARR growth is ahead of expectations, reaching USD1bn in July 2026; the 1GW data center, breakthroughs in inference using domestic chips, and the acquisition of XCore Sigma should help alleviate computing constraints; subsequent GLM iterations could provide positive catalysts.

  Weaknesses

  A shift in the revenue mix toward low-margin API business could dilute margins; long-term FCF assumptions have been revised downward; the July 2026 H-share issuance increased the share count.

  Comparison

  The report says the target price implies a December 2027 P/ARR of 14x, higher than Anthropic's 8x. The premium is justified by Z.AI's ARR growth of approximately 220%, compared with approximately 50% for Anthropic.

  Risks

  Competitor model iterations, potential restrictions in the US market, the expiration of the lock-up period in January 2027, and competitor IPOs or financings could weaken the scarcity premium.

• Kimi K3

  Competitive and technological implications

  Strengths

  K3's parameter scale increased to 2.8trn, which the report views as evidence that the scaling law remains effective.

  Weaknesses

  Following the K3 launch, new-to-C subscriptions were suspended due to slow token throughput, indicating that computing supply may be insufficient.

  Comparison

  The launch of K3 increases competitive pressure on Chinese open-weight models such as Z.AI, MiniMax, and DeepSeek to release larger-parameter models in the third quarter of 2026.

  Risks

  Rapid iteration by competitors such as Kimi could dilute Z.AI's token usage growth and valuation scarcity.

Key data

• RatingHoldHSBC maintains its rating.
• Target priceHKD1,500.00Lowered from HKD1,900.00.
• Current share priceHKD1,281.00As of July 27, 2026.
• Implied upside+17.1%Calculated based on the target price and current share price.
• December 2026 ARR forecastUSD2bnRaised from USD1bn.
• December 2027 ARR forecastUSD6.5bnRaised from USD2.4bn.
• Profitability inflection point2027HSBC expects Z.AI to turn profitable in 2027, earlier than the previous expectation of 2028.
• Key DCF assumptionsWACC 11.8%, terminal growth rate 3%Used to calculate the target price.
• 2027 P/ARR14xHSBC says the target price implies a December 2027 P/ARR of 14x, higher than Anthropic's 8x, but Z.AI has faster ARR growth.

Impact & implications

For investors, Z.AI's commercialization and computing reserves provide growth support, but shorter technology leadership cycles in the AI model industry, intensifying price competition, and potential increases in IPO supply will weigh on the scarcity premium. The Hold rating reflects the balance between upgraded growth expectations and valuation pressure.

Risks

• New model launches by competitors could dilute Z.AI's token usage growth.
• Intensifying price competition could depress long-term margins and FCF.
• Chinese frontier models face geopolitical risks, including potential bans in the US market.
• Approximately 38% of Z.AI's shares may be released from lock-up in early January 2027.
• Financing or IPOs by frontier labs such as Anthropic, OpenAI, and Kimi could weaken Z.AI's scarcity premium.
• Subsequent model iterations may perform below expectations.

What to watch

• Progress on the launch of the new GLM model in the third quarter of 2026.
• The model iteration pace of Kimi, MiniMax, DeepSeek, and major US AI companies.
• Trends in Z.AI's OpenRouter token usage and growth in API demand.
• The actual improvement in computing utilization from the 1GW data center, domestic-chip inference, and XCore Sigma integration.
• Share supply pressure following the expiration of the lock-up period in January 2027.
• The impact of financing and IPOs by frontier AI companies on the valuation scarcity premium.