Wall Street reviews Zhipu's earnings report: The explosive ARR is the biggest surprise, and the simultaneous rise in token quantity and price indicates that there is already substantial pricing power

Ask AI · In the context of price wars, how does the simultaneous rise in token volume and price reflect pricing power?

Zhipu AI has delivered a performance report that has caught Wall Street’s attention, with annual revenue more than doubling year-over-year, but what truly shocked the market was a real-time data disclosed after the earnings release:

As of March 31, 2026, the annual recurring revenue (ARR) of the company’s open platform API has soared to approximately 1.7 billion RMB (about $250 million), a more than 2.4-fold increase from about 500 million RMB at the end of 2025, and a 60-fold increase compared to 12 months ago. Both Morgan Stanley and J.P. Morgan regard this as the biggest surprise of the earnings report. More convincingly, this growth was not simply achieved through “price to volume” trade-offs.

According to Chasing Wind Trading Platform, J.P. Morgan specifically pointed out in its research report that, the token price on Zhipu’s API platform has increased by 83% since the beginning of the year, while demand continues to accelerate — volume and price are rising together. In the current environment of intense domestic large-model price wars, this phenomenon is extremely rare and directly confirms that Zhipu has established substantial pricing power in high-value scenarios such as programming and intelligent agents.

From the profit path perspective, Zhipu’s loss structure is undergoing a qualitative change. In 2025, the company’s R&D expenditure was about 3.2 billion RMB, roughly equal to the adjusted net loss scale, which means the gross profit generated by the existing models is sufficient to cover sales and administrative expenses — core business has already achieved breakeven at the contribution level, and all losses are essentially proactive investments in next-generation model iterations.

J.P. Morgan believes that, with the continued triple-digit revenue growth and the sustained expansion of API gross margin (only 3% in 2024, rising to 19% in 2025), the profitability timeline is becoming clearer, and the company is expected to turn profitable by 2029.

ARR Explosion: From “Year-End Goal” to “Already on Track”

The most core highlight of this earnings report is the ARR data disclosed by Zhipu. As of March 31, 2026, the ARR of the open platform API has reached about $250 million, a 6.4-fold increase from the beginning of the year, and about 60 times higher than 12 months ago.

Management’s year-end target is $1 billion, and current progress shows that this goal is not an unreachable vision but is already on a rapid fulfillment track.

Morgan Stanley listed the outperformance of ARR as a “strengthening investment logic” in its research report and characterized it as a “Major Surprise.”

From a business structure perspective, cloud deployment revenue in the second half of 2025 grew by 431% year-over-year, far exceeding the 57% growth of private deployment, and the proportion of cloud business in total revenue quickly increased from a single-digit percentage to 26%. This structural shift indicates that Zhipu’s business model is evolving from asset-heavy, low-repeat project-based delivery to a light-asset, high-stickiness subscription API economy.

Token volume and price rising together: Pricing power is the most scarce signal

In the current environment where domestic large-model tracks are generally caught in “price competition,” Zhipu’s ability to achieve an 83% increase in token price since the beginning of the year while demand remains strong warrants in-depth analysis.

Olivia Xu, an analyst at J.P. Morgan, explicitly pointed out in the research report that volume and price rising together is the clearest signal driven by “model real competitiveness” and “high-value workload” growth.

Specifically, for customers involved in programming (Coding) and intelligent agent (Agent) scenarios, the payment logic has upgraded from “pay-per-volume” to “pay for task completion quality, throughput, and stability” — an inherently higher-dimensional business relationship. When customers are willing to pay a premium for better results rather than just seeking the lowest unit price, pricing power has quietly formed.

From the perspective of model iteration paths, Zhipu’s rapid evolution from GLM-4.5/4.6/4.7 to GLM-5, along with continuous investments in production-level programming, long-context reasoning, and multi-step execution stability, is the technical foundation supporting this pricing power.

Gross Margin Turning Point: From Loss to Profit in Cloud Business

In the second half of 2025, Zhipu’s cloud deployment gross margin jumped significantly from -0.4% in the first half to 22.4%, marking the cloud business’s official crossing of the profit threshold and entering a positive cycle driven by scale effects.

Looking at the overall group, the gross margin of the open platform API increased from 3% in 2024 to 19% in 2025, a 16 percentage point rise.

J.P. Morgan expects that, with scale continuing to expand and inference efficiency improving, there is still considerable room for gross margin to grow. According to its forecast model, the overall gross margin of the group will remain around 31% in 2026, further rising to 36% in 2027, and reaching 37% in 2028.

It is worth noting that the gross profit for the full year of 2025 was about 297 million RMB, while sales and administrative expenses for the same period totaled approximately 896 million RMB, and R&D expenditure was about 3.2 billion RMB. Excluding R&D investments, gross profit has basically covered non-R&D operating costs, indicating that Zhipu’s core business model already has self-sustaining capacity. The current losses are purely strategic R&D investments rather than a flaw in the business model itself.

Private Deployment: Upgrading Potential of Existing Assets

In addition to the explosive growth of cloud API business, Zhipu’s accumulated private deployment foundation in regulated industries in China is also a key pillar of J.P. Morgan’s investment logic. In the second half of 2025, private deployment revenue reached 372 million RMB, a 57% increase year-over-year, and the absolute scale is still more than twice that of the cloud business.

J.P. Morgan believes that this large existing customer base has unique strategic value: as foundational models continue to iterate and upgrade, these deployed customers have natural upgrade needs, which could evolve into cyclical, predictable upgrade-driven revenue.

Highly sensitive data security requirements in regulated industries such as finance, government, and energy also make private deployment an irreplaceable delivery mode in the foreseeable future, forming a structural moat that differentiates Zhipu from pure cloud competitors.

J.P. Morgan Significantly Raises Target Price

J.P. Morgan maintains an “Overweight” rating, raising the target price from HKD 800 to HKD 950, based on a 30x expected P/E ratio for 2030, discounted to the end of 2026 at a 15% weighted average cost of capital. The 30x valuation premium mainly reflects the company’s expected over 100% compound annual revenue growth from 2026 to 2030.

According to the latest forecasts, Zhipu’s revenue in 2026 will reach 3.19B RMB (a 341% YoY increase), further rising to 7.26B RMB in 2027, 19.725 billion RMB in 2028, and potentially surpassing 98.8 billion RMB by 2030. Adjusted net profit is expected to turn positive by 2029 at 19.73B RMB, further increasing to 20.36 billion RMB in 2030.

Morgan Stanley also maintains an “Overweight” rating, with a target price of HKD 560, using DCF valuation with assumptions of a 15% WACC and 3% perpetual growth rate, corresponding to a 53x sales multiple in 2027. While the methodologies differ, both institutions share a highly consistent outlook on Zhipu’s long-term value.