Alibaba Is in Talks to Lead $300M UniPat Round at $2.5B Valuation
The prospective investment would put Alibaba, Tencent and HSG behind a young startup focused on synthetic training data and AI evaluation. The round is expected to close soon, but its terms remain unsettled.
Listen to this story
The audio brief
Story brief
3 key pointsUniPat AI is negotiating a proposed $300 million financing that would value the young synthetic-data and evaluation startup at $2.5 billion, with Alibaba expected to lead and Tencent and HSG potentially joining. The deal has not closed, so valuation, size and investor lineup may change. UniPat’s scope covers training data and tests for coding agents, browser automation and visual reasoning, plus lightweight...
- 01
Founded in late 2025 by Li Kuan, a former intern at Alibaba’s Tongyi AI lab.
- 02
Monolith and ByteDance-backed Jinqiu Fund are listed as early UniPat investors.
- 03
Negotiated terms—not a completed investment—leave closing and final economics uncertain.
Alibaba is reportedly in talks to lead a $300 million financing round for UniPat AI, a startup that generates synthetic training data and designs evaluation scenarios. The proposed deal would value UniPat at $2.5 billion; Tencent and HSG are expected to participate, but talks remain active.
The financing has not closed, and final terms could change. The reported round size, valuation and investor lineup therefore remain proposed. Representatives for Alibaba, Tencent, HSG and UniPat did not immediately respond to Bloomberg’s requests for comment.
A company built around data and tests
UniPat’s reported work combines synthetic training data with evaluation scenarios for several kinds of AI systems. Its focus spans coding agents, web-browser automation and visual-reasoning models, placing both model development material and model assessment within the company’s stated remit.
UniPat’s reported focus
- Synthetic training data generation.
- Evaluation scenarios for coding agents and web-browser automation.
- Evaluation scenarios for visual-reasoning models.
A young startup with a wider research remit
UniPat was founded in late 2025 by Li Kuan, a former intern at Alibaba’s Tongyi AI lab. Alongside its data and evaluation work, UniPat’s research lab develops lightweight models for scientific research and predictive applications. That makes the company’s reported activity broader than evaluation work alone.
Monolith and ByteDance-backed Jinqiu Fund are listed as early UniPat investors. The company operates in a broader market that includes U.S. data-annotation and benchmarking businesses such as Scale AI, Mercor and Artificial Analysis. If negotiations conclude on the reported terms, Alibaba, Tencent and HSG would join that investor base.
The final terms remain the decision point
The reported valuation is tied to a transaction still being negotiated, not a completed investment. A closing would confirm whether UniPat receives the stated $300 million and whether Alibaba leads a group that includes Tencent and HSG. Until then, the financing is a significant proposed vote of support, with its central commercial terms still subject to change.
Sources
- seekingalpha.comAlibaba reportedly slated to lead $300M round in AI startup UniPat at $2.5B valuation
- longbridge.com[AI] Alibaba reportedly plans to lead the investment in AI startup UniPat AI
Loading discussion...
Reader comments
Newest comments first. Replies stay oldest first.