J.P. Morgan maintains an Overweight rating on Zhipu AI and raises its target price to HK$950

Overview

This is a company research report by J.P. Morgan on Knowledge Atlas Technology Joint Stock Co. Ltd. (Zhipu AI). The report's core view is that Zhipu AI is showing a steady upward trend in model capability iteration, open-platform API demand, token consumption intensity and API pricing, with commercialization quality better than previously expected. The analyst therefore raises 2026-2030E revenue forecasts and maintains an Overweight rating.

Core views

The report believes Zhipu AI has reached an important inflection point, especially in its global API business. The releases from GLM-4.5/4.6/4.7 to GLM5, as well as the strategic shift toward agent systems, tool-enhanced reasoning, developer infrastructure and multi-step execution stability, have increased its commercial value in coding and agent-related scenarios. Open-platform API gross margin rose from 3% in 2024 to 19% in 2025; if triple-digit revenue growth and API margin expansion persist, the path to narrowing losses will become more visible.

Analysis framework

The report analyzes API ARR, token prices, revenue forecast revisions, gross margin changes, R&D spending, cash-basis SG&A coverage and a long-term P/E valuation framework. Its investment thesis is not just about near-term losses, but about whether the company can continue to remain in the global top-tier foundation model capability cohort and convert that capability into sustainable revenue from APIs and deployment businesses.

Methodology notes

• Industry competitiveness analysisFoundational model capability ladder framework

  Model capability determines long-term economic outcomes

  The report argues that in the foundation model industry, long-term value is mainly determined by whether a company can maintain globally top-tier model capability across multiple technology cycles; the business model, deployment format and near-term profit margins are downstream manifestations of that capability.

• Valuation methodsForward P/E discount valuation

  30x 2030E P/E discounted back to Dec-26

  The HK$950 target price comes from around Rmb43 in adjusted EPS for 2030E, a 30x target multiple, a 15% WACC, and an assumed USD/HKD exchange rate of 7.8.

• Operating quality analysisAPI commercialization quality framework

  rising token volume and price together

  The report treats the growth in token demand and the 83% year-to-date increase in token prices as evidence of model differentiation, stronger task completion quality and improved customer willingness to pay.

Asset mapping & comparison

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

• Knowledge Atlas Technology Joint Stock Co. Ltd. (Zhipu AI) / 2513.HK

  Covered company; J.P. Morgan maintains an Overweight rating

  Strengths

  Rapid growth in API ARR, rising token volume and prices, ongoing GLM model iteration, strong commercialization potential in coding and agent scenarios, and an existing local deployment base in regulated Chinese industries.

  Weaknesses

  The company is still loss-making, R&D spending is high, adjusted net profit remains negative for 2026-2028E, and the profitability path depends on strong revenue growth and continued improvement in API gross margins.

  Comparison

  A 30x 2030E P/E implies a valuation premium to leading Chinese internet companies, mainly reflecting the assumption of more than 100% revenue CAGR over 2026-2030E.

  Risks

  Export controls, geopolitical risk, entity list designation risk, intensifying competition, reliance on compute infrastructure and external suppliers, and uncertainty around commercialization and customer adoption.

Key data

• API platform ARRUS$250mnAs of March 31, 2026, it is up 6.4x year to date and 60x over the past 12 months; management's year-end target is US$1bn.
• Token price change+83% YTDPrices and demand are rising together, which is seen as a core signal of improved commercialization quality.
• 2026-2030E revenue forecast increase46%-78%J.P. Morgan raises revenue forecasts for the next five years, mainly driven by expanding demand for the open-platform API.
• 2025 R&D investmentRmb3.2bnThis amount is roughly equal to adjusted net loss, leading the report to conclude that gross profit from existing models already covers SG&A on a cash basis.
• Open platform API gross margin2024年3%升至2025年19%Reflects improvements in scale, utilization and model efficiency.
• 2030E revenueRmb98,832mnThe new forecast is 47% above the previous forecast of Rmb67,063mn.
• 2030E adjusted net profitRmb20,360mnUp 19% from the previous forecast of Rmb17,146mn.
• Dec-26 target priceHK$950Previous target price was HK$800; Overweight rating maintained.

Impact & implications

If the report's judgment is correct, Zhipu AI's investment story will shift from being a large-model company with simply high R&D spending and high losses to a growth-oriented AI platform with model leadership, rising API volume and pricing, gross margin expansion and a long-term path to profitability. For investors, the key variable is whether API growth can continue to approach management's target and whether high R&D spending can keep translating into model leadership and commercialization premium.

Risks

• Export controls, geopolitical risks and possible entity list designation could affect model R&D, the supply chain or customer adoption.
• Intensifying competition in the foundation model industry could pressure pricing, raise customer acquisition costs or weaken model differentiation.
• Continued high R&D spending creates execution risk and may keep weighing on profitability.
• Commercialization and customer adoption remain uncertain, especially whether API demand can continue advancing toward management's US$1bn year-end ARR target.
• Dependence on compute infrastructure and external suppliers brings cost and availability risks.

What to watch

• Whether API platform ARR can continue to move from US$250mn toward management's year-end target of US$1bn.
• After token prices rose 83%, whether demand can still keep accelerating to validate the sustainability of higher volume and price.
• Whether open-platform API gross margin can continue to expand from the 19% level in 2025.
• Whether model iteration from GLM-4.5/4.6/4.7 to GLM5 can sustain leadership in coding, agent and long-context reasoning scenarios.
• Whether the 2029 profitability expectation and the 2030E Rmb20.36bn adjusted net profit forecast are revised further up or down.