The Real Bottleneck in the Age of AI Agents Isn't the Model -- It's the Data

News provided by

09 Jun, 2026, 10:00 CST

Singdata Closes Series B at over USD $100M, Launches in Hong Kong with a Bet on "AI-Native Data Infrastructure" — Aligned with the HKSAR Government's HKD 3 Billion "AI+" Strategy

HONG KONG, June 9, 2026 /PRNewswire/ -- Enterprises worldwide are racing to deploy AI Agents. Yet according to 2026 reports from Deloitte and Google Cloud, only 31% have reached production. 46% of organizations cite systems integration as the top barrier, and 42% point to data access and quality. The bottleneck is no longer the model — it is whether the underlying data is AI-Ready.

Singdata (Yunqi Tech), an AI-native data infrastructure company, today announced from Hong Kong the close of its Series B funding round, bringing total funding to over USD $100 million. The milestone marks an inflection point for the Asia-Pacific "Data + AI" infrastructure category — a category where the international comparable, Databricks, has seen its valuation grow 3.1x in the past 18 months to USD $134 billion.

Why Hong Kong, Why Now — After Compute, Data Is the Next Battlefield

The HKSAR Government is driving a HKD 3 billion "AI+" strategy: a HKD 1 billion Hong Kong AI Research and Development Institute (HKAIRDI), the Sandy Ridge supercomputing centre breaking ground in 2026, and citywide compute capacity expanding 36-fold by 2032. Hong Kong is positioning itself as a dual data hub between mainland China and the international market.

"Once compute becomes abundant, the next bottleneck is the data layer that powers AI applications." Singdata's Hong Kong launch is a direct response to that thesis — when enterprises already have compute and mature models, what determines whether AI works in production is a paradigm upgrade in data infrastructure. Singdata's recently published technical thesis, Five Design Principles of an AI-Native Data Platform, sets out the technical path and design principles behind this paradigm shift in full.

Solving Two Structural Bottlenecks for Enterprise AI

Singdata's core innovations — Generic Incremental Compute (GIC) and a Lakehouse architecture — address the two structural pain points of enterprise AI deployment:

AI-Ready data: Native support for documents, audio, video, chat logs and other unstructured "dark data" that constitutes over 80% of enterprise data — automatically parsed, vectorized, and aligned with LLM inference.

Real-time, low-cost retrieval: GIC processes only the data that changes, decisively breaking the cost-and-latency ceiling of traditional Lambda architectures' "full-rerun" approach, and supporting the high-frequency, millisecond-level vector search that AI Agents require.

Singdata's AI Lakehouse is compatible with all major cloud platforms, and Singdata is a major contributor to the Apache Iceberg open standard, ensuring customers retain 100% sovereignty and portability over their own data — strategically significant in an Asia-Pacific market where data sovereignty is fast becoming a board-level concern.

Customers: Cross-Industry, Cross-Geography Enterprise Validation

Singdata's customer base spans Asia's leading enterprises:

Internet: AntGroup, Rednote, Kuaishou

Automotive: Changan Auto, Great Wall Motor, Toyota, Volkswagen

Logistics & Commerce: Ninja Van, Synagie, Atlas

The breadth of industries and geographies served is, in itself, the most direct proof that Singdata's architecture adapts to highly differentiated enterprise environments.

Team: Alibaba "Apsara" Veterans, Combining Global Vision with Asia-Pacific Execution

Singdata's founding team draws on senior alumni from Microsoft, Oracle, Alibaba Cloud, and ByteDance's Volcano Engine — and was central to building Alibaba Cloud's "Apsara" data platform, one of the largest cloud-native data systems globally. The team combines deep familiarity with global technical frontiers and large-scale Asia-Pacific operational experience.

Founder Statement

"The data stack enterprises built over the past two decades was designed for human analysts reading T+1 dashboards. An AI Agent can fire ten thousand queries a second. Run that workload on the old architecture and every Agent thought triggers an expensive full table scan — that is the real reason enterprise AI is failing to scale. What Singdata is rebuilding isn't just the engine — it's the paradigm: a data platform designed for Agents, not only for analysts."

— Yu Sicheng, CEO, Singdata (Yunqi Tech)

This funding round will be fully invested in continued R&D of Singdata's AI-native data platform, accelerating its trajectory to become the platform of choice for Data + AI infrastructure across Asia-Pacific, with Hong Kong serving as its strategic gateway to international markets.

About Singdata (Yunqi Tech)

Founded in 2022, Singdata is an enterprise AI-native data infrastructure company. Its proprietary AI Lakehouse, powered by Generic Incremental Compute (GIC), bridges the last mile between enterprise data and AI — already trusted by leading organizations across Asia-Pacific including Rednote, Kuaishou, Toyota, Volkswagen, and Ninja Van. Singdata is a core contributor to the Apache Iceberg open standard and is committed to becoming the leading Data + AI infrastructure platform for Asia-Pacific.

Further Reading | Full Technical Thesis

Five Design Principles of an AI-Native Data Platform — Singdata's systematic thesis on the paradigm shift in data infrastructure for the AI era, covering unified Lakehouse storage, AI as a native compute engine, the Medallion Architecture with incremental compute, Agent-friendly development paradigms, and enterprise-grade governance:
https://www.yunqi.tech/resource/blogs/ai-native-data-platform

Product demo & consultation: https://www.yunqi.tech/reservation

SOURCE Singdata