Liepin (TGDLF): AI Product Cash Billings Nearly Double in 1H 2026

Liepin (TGDLF) said on its August 23, 2026 earnings call that cash billings from its AI products grew nearly 100% year over year in the first half of 2026, building on a base that had already exceeded RMB 100 million for full-year 2025 [1]. The company has not disclosed AI revenue, an AI segment, or AI as a share of total revenue, so the growth rate cannot be tied to any audited line item.

What Liepin sells

Liepin matches mid- to high-end talent with employers in China. Corporate clients pay an annual fee to open an account, then post jobs, search the resume database and contact candidates directly; Liepin charges by account tier and by value-added service, while the consumer side contributes a smaller stream from career training and development.

The business had been shrinking. Revenue over the last four full years ran RMB 2,638 million, 2,282 million, 2,081 million and 1,986 million, three straight annual declines. The first half of 2026 was the first growing half in that stretch, with revenue of RMB 989 million, up 5.6%, including RMB 820 million of enterprise recruitment revenue, up 6.7% [2].

Where AI enters the product

AI sits in two specific places. The first is the account ladder: an AI Pro tier priced above the existing AI account, which clients pay more to upgrade into. The second is Intended Candidate Search, which screens the resume database against a job description for people likely to be interested, and for most roles returns the first candidate who expresses interest within one hour.

Both run on Liepin's in-house large language model, Tongdao Hui Cai, which uses a closed-loop reinforcement learning system that keeps training on real feedback from the hiring process. In the first half, AI and AI Pro accounts together covered close to 100% of enterprise clients. About 40% of AI-account clients upgraded to AI Pro within six months of launch, more than 70% of new clients took AI Pro at signing, and average daily orders and resume downloads each grew more than 70% year over year.

The revenue evidence

Management disclosed the upsell through one metric only. Cash billings from AI products exceeded RMB 100 million for full-year 2025 and grew nearly 100% year over year in the first half of 2026 [1].

Placed back in the revenue structure, paid enterprise customers stood at roughly 63,000, up only 0.4% year over year, so the customer count barely moved [2]. The additional money came from the same clients buying a more expensive account rather than from new clients arriving. On management's own figures, the AI cash-billings increment was roughly RMB 40 million to 45 million, against RMB 52 million of total revenue growth [2].

The cost evidence

On the cost side the change shows up in gross margin and R&D. Consolidated gross margin was 79% in the first half, up 1.4 percentage points year over year and the highest in four years. R&D expense was RMB 109 million, down 13.3%, and the R&D expense ratio fell to 11%, down 2.4 points [3].

Management attributed the margin gain to AI product mix and higher delivery-team productivity, and attributed the R&D decline both to internal adoption of AI tools and to a reorganization of the R&D function. It guided full-year gross margin to hold the first-half trend and expects the sales and R&D expense ratios to keep falling [3].

What the numbers do not prove

Each figure has a boundary. Cash billings is a collection metric, not a revenue line, and the prior-year base behind the "nearly 100%" was never given. A 79% gross margin is below the 79.4% the company earned before the downturn, so a four-year high is equally consistent with reversion to its earlier range. R&D expense has fallen for four consecutive years, at RMB 400.3 million, 360.4 million, 303.8 million and 272.1 million, a decline that was underway before AI was offered as the explanation [3].

The hiring market is also recovering. The company reports new job postings up 14% year over year, and a market recovery raises spending per client through the same channel an AI upgrade tier does [2]. One further limit comes from the data: Drillr's financial statements for this issuer stop at December 2025, so every first-half 2026 figure is management's own. Only the prior-year base can be checked independently, and on that base revenue, gross margin, R&D and net profit all reconcile [2].

What is confirmed and what is not

Liepin returned to revenue growth with a paid customer count that barely changed, and management assigns most of that growth, along with the gross margin and R&D improvements, to its AI products. This is the third reporting period with a disclosed AI billings figure, moving from RMB 10 million to more than RMB 100 million to roughly double again, so the series itself is complete [1].

What remains open is what AI is worth on the income statement. There is no AI revenue disclosure, no headcount series to corroborate the efficiency gain, and two competing explanations, the reorganization and the cyclical hiring recovery, that still stand.

Use case: revenue growth and cost reduction. Adoption stage: mature. Value released: significant. Confidence: medium.

Sources

[1] Drillr · Liepin (TGDLF) · 2026-08-23 · earnings call

"In terms of cash billings from our AI products, it exceeded RMB 100 million for the full year last year. In the first half of this year, AI products maintained a strong momentum with cash billings increasing by nearly 100% year-on-year."

[2] Drillr · Liepin (TGDLF) · revenue trend verification

[3] Drillr · Liepin (TGDLF) · profit and expense verification