ZhiPu (02513.HK) Releases First Annual Report Since Listing: Revenue Doubles to RMB 724 million, Net Loss Widens to RMB 4.7 billion

ZC Asia has learnt that on 1 April 2026, ZhiPu (02513.HK), the ‘first large-model stock’ on the Hong Kong Stock Exchange, released its first annual financial report since listing. On the previous trading day, buoyed by the results, the company’s share price surged by over 30%, hitting an all-time high during trading, with its total market capitalisation briefly exceeding HK$400 billion.

APIs and Agents Emerge as Dual Growth Engines

In 2025, ZhiPu achieved total revenue of RMB 724 million, representing a year-on-year increase of 131.9%, a growth rate far exceeding market expectations. In terms of revenue composition, growth was not driven by a single business segment but rather demonstrated a clear ‘dual-engine’ structure.

Revenue from APIs and the open platform reached RMB 190 million, a year-on-year increase of 292.6%, with its share of total revenue rising from 14.6% in 2024 to 26.3%, making it the fastest-growing business segment. In a research report published following the financial results, JPMorgan noted that by the end of March 2026, annual recurring revenue for this business had surged to US$250 million, whilst the price of API tokens had risen by 83% year-to-date, demonstrating a trend of “rising volume and price”. This is viewed by institutions as a structural shift driven by genuine market demand.

Revenue from enterprise-grade AI agents reached 166 million yuan, a year-on-year increase of 248.8%, with its share of total revenue rising to 22.9%. As a key form of AI commercialisation, the rapid growth of the AI agent business indicates that Zhipu has established a replicable delivery capability in the enterprise services sector, with its commercialisation strategy having been preliminarily validated. The combined revenue share of these two business segments has approached 50%, signalling that Zhipu is accelerating its transition from the ‘model capability demonstration’ phase to the ‘scalable commercialisation’ phase.

Widening Losses and Pressure on Gross Margin

In 2025, Zhipu recorded a net loss of 4.718 billion yuan for the year, a 59.5% year-on-year increase from the 2.957 billion yuan loss in 2024. The primary cause of this widening loss was research and development (R&D) expenditure, which reached 3.18 billion yuan—a 44.9% year-on-year increase, equivalent to 4.4 times total revenue. In terms of allocation, R&D expenditure was primarily directed towards model iteration, computing infrastructure development and the expansion of the algorithm team. Against the backdrop of an ongoing ‘arms race’ in the large-scale model sector, this strategy of ‘trading losses for technological leadership’ comes as no surprise; however, the key question is whether R&D investment can be consistently converted into commercial outputs with high gross margins.

It is worth noting that, despite a doubling of revenue, Zhipu’s overall gross margin fell from 56.3% in 2024 to 41.0%, a decline of 15.3 percentage points. The company explained in its financial report that the decline in gross margin was primarily driven by the increased proportion of low-margin cloud-based business. This reflects a practical issue within Zhipu’s current growth structure: whilst high-growth businesses are making a greater contribution to revenue, they have not yet generated commensurate profit quality. Should this trend persist, there may be a temporary disconnect between future revenue growth and profit improvement.

Potential risks should not be overlooked

Beyond the impressive growth figures, Zhipu’s potential risks also warrant careful assessment by investors.

Firstly, there is the issue of the sustainability of Token pricing. Whilst a JPMorgan report offered a positive assessment of the “double-digit growth in both volume and price” of APIs, Chairman Liu Debing admitted during the earnings conference call that Tokens would move towards a tiered structure in the future. High-tier Tokens, characterised by high complexity and reliability, will retain pricing power, but price pressure on mid- and low-tier Tokens is inevitable. This implies that the sustainability of Zhipu’s current high-pricing strategy hinges on its ability to maintain a technological lead in complex task scenarios, as well as on clients’ actual perception of the return on investment. Should competitors develop the capability to offer alternatives in equivalent scenarios, Zhipu’s pricing power will be put to the test.

Secondly, there is the issue of cash flow and reliance on financing. An annual loss of 4.7 billion yuan, coupled with sustained, high-intensity R&D investment, places continuous pressure on Zhipu’s cash flow. Although Zhipu, as the “first large-model stock” listed on the Hong Kong Stock Exchange, enjoys certain financing advantages in the capital market, if the pace of commercialisation fails to match the rhythm of R&D expenditure, it may still face the dual constraints of financing needs and equity dilution in the future.

Furthermore, the competitive landscape within the industry is becoming increasingly intense. The AI large-model sector currently remains in a phase of “multiple strong players coexisting”, with leading domestic and international firms continuing to ramp up their investments in model capabilities, ecosystem development and client resources. Although Zhipu has secured a first-mover advantage in its API and intelligent agent businesses, whether it can maintain its market share leadership across a broader range of industry scenarios remains to be seen.

Institutions are optimistic about long-term growth, whilst short-term valuations remain subject to market speculation

Faced with this “mixed bag” of financial results, investment banks’ views have diverged, though leading institutions continue to send positive signals. Following the release of the results, JPMorgan maintained its “Overweight” rating and raised its target price to HK$950. The core rationale lies in the fact that strong API demand, sustained growth in annual recurring revenue and rising token prices collectively point to growth driven by genuine demand, rather than short-term marketing stimuli.

Zhang Peng, CEO of Zhispu, further emphasised at the earnings conference that the growth in API revenue marks the beginning of a structural, long-term trend. As AI begins to solve complex problems, the invocation and consumption of tokens will translate into tangible economic value, with the potential for exponential growth in the future.

Disclaimer: The content herein is for reference only and does not constitute any form of investment advice. Should any information contained herein prove inaccurate, incomplete, or potentially misleading, please refer to the company's official announcements. Market risks exist; investment requires caution.