Building the Data Infrastructure for Physical AI: Hesai Launches Kosmo 3D World Data Engine

Hesai launches Kosmo, promising 99% robot task success and 80% production cost savings.

What the company is saying

Hesai Group is positioning Kosmo as a breakthrough in Physical AI infrastructure, emphasizing its ability to deliver high-quality real-world 3D data at scale for robotics, world modeling, and immersive content. The company highlights Kosmo One, described as the world's first 8K AI spatial camera, and the Kosmo World platform, which together enable rapid, detailed scanning and reconstruction of physical spaces. Strategic partnerships with Galbot, Dexmal, Pixar, and Amap are showcased to demonstrate Kosmo's versatility across robotics, film, and mapping. The announcement features concrete operational metrics, such as reducing 3D training environment creation from 300 hours to 15 minutes and achieving a 99% robot task success rate in a Dairy Queen deployment. Hesai's executives, Co-Founder and CTO Shaoqing Xiang and CEO Dr. Yifan "David" Li, frame Kosmo as the next growth driver, building on a decade of lidar expertise and a leading position in ADAS and robotaxis. The company also stresses its open ecosystem approach and dual business models, targeting both integrated solutions and large-scale data services. The tone is assertive, with repeated claims of industry firsts, efficiency multiples, and market leadership, but without supporting financial figures.

What the data suggests

The disclosed numbers show Kosmo can scan a 100-square-meter indoor space in about 15 minutes, with Kosmo World processing data up to five times faster than competitors. In a real-world Dairy Queen deployment, Kosmo reduced 3D training environment creation time from 300 hours to 15 minutes, enabling the Sharpa robot to achieve a 99% task success rate and cut failure rates to one-tenth of previous levels. Kosmo can generate over 10,000 training trajectories from a single space, delivering more than 100 times the training data for the same cost. In film production, Kosmo enabled a 10,000-square-meter streetscape to be digitized in one week instead of two months, with up to 80% cost savings. Hesai's actuation modules, including the Sharpa Wave hand, deliver three times the power density of peers in a package 37% smaller, with cumulative deliveries exceeding 10,000 units by Q2 2026 and six-figure shipments targeted for 2027. Amap integration brings Kosmo's 3D environments to nearly one billion monthly active users. While these metrics are operationally impressive, there are no revenue, margin, or profit figures, and claims of global market leadership are not backed by delivery or share data.

Analysis

The announcement is upbeat and rich in technical and operational detail, highlighting the launch of Kosmo and its integration with Kosmo One and Kosmo World. Several claims are substantiated with specific, realised case studies (e.g., Dairy Queen deployment, film/media scanning, and actuation module deliveries exceeding 10,000 units by Q2 2026). However, the release lacks any financial metrics—no revenue, profit, or cash flow figures are disclosed—so the investment impact cannot be fully assessed. Some claims, such as 'shipments are expected to reach six figures in 2027' and 'the aim is to extend navigation from finding a store to locating a product inside it,' are forward-looking but limited in number. The tone is somewhat inflated by repeated references to 'world's first,' 'leading global market position,' and large efficiency multiples, but these are partially supported by operational examples. The absence of financial data and market share specifics means the true signal cannot exceed weak_positive, and the hype level is moderate due to the ambitious language.

Risk flags

• ●The absence of financial disclosures such as revenue, profit, or cash flow tied to Kosmo or overall operations makes it difficult to assess the commercial impact or profitability of these technological advances.
• ●Claims of global market leadership and efficiency multiples are not substantiated with independent benchmarks, delivery numbers, or market share data, raising the risk that competitive positioning is overstated.
• ●Execution risk remains around scaling Kosmo's adoption from pilot deployments to widespread commercial use, especially as six-figure shipment targets for actuation modules are forward-looking and not yet realized.
• ●Reliance on high-profile partnerships (e.g., Amap, Pixar, NVIDIA) for ecosystem growth introduces dependency risk, as continued collaboration and integration are not contractually guaranteed.
• ●The ambitious efficiency and cost-saving claims (e.g., 80% reduction, 100x data efficiency) may not generalize across all use cases or customer segments, and actual realized savings could vary materially.

Bottom line

Hesai's Kosmo launch is backed by concrete operational metrics—15-minute scans, 99% robot task success, and 80% production cost savings—demonstrated in real deployments and supported by partnerships with major players in robotics, mapping, and film. The company is targeting rapid growth, with actuation module deliveries already exceeding 10,000 units and six-figure shipments expected in 2027. However, the lack of financial disclosures and unsubstantiated market leadership claims mean investors cannot yet assess the revenue or profit impact of these advances. The narrative is credible at the technical level but requires future updates with financial results and broader commercial adoption to confirm its investment case. The most important takeaway is that Kosmo's operational potential is proven in pilot settings, but the scale and profitability of its impact remain to be demonstrated.

Announcement summary

(NASDAQ:HSAI, HKEX:2525) Hesai Group announced the launch of Kosmo, a 3D world data engine designed to address the challenge of accessing high-quality real-world data at scale for Physical AI applications. Kosmo integrates Kosmo One, described as the world's first 8K ultra-high-definition AI spatial camera, with the Kosmo World cloud platform to transform physical environments into detailed 3D digital spaces for robot training, world model development, and immersive content creation. At the launch event, Hesai announced strategic partnerships with embodied AI companies Galbot and Dexmal, and demonstrated Kosmo's applications with Pixar's RenderMan and Amap. Kosmo One features high-resolution lidar, a one-inch imaging sensor, and a high-precision inertial measurement unit, enabling microsecond-level synchronization and millimeter-level geometric accuracy. A single operator can scan a typical 100-square-meter indoor space in approximately 15 minutes. Kosmo World processes captured data into high-quality 3D models at speeds up to approximately five times faster than comparable products. In a Dairy Queen ice cream store deployment, Kosmo reduced the time to create a 3D training environment from approximately 300 hours of manual modeling to just 15 minutes, resulting in the Sharpa robot's task success rate reaching 99% and the failure rate dropping to one-tenth of previous levels. Kosmo can automatically generate more than 10,000 training trajectories from a single captured space, providing over 100 times as much high-quality training data for the same cost. In film and media, Kosmo enabled a 10,000-square-meter European streetscape to be scanned, reconstructed, and edited in one week instead of two months, with potential cost savings of up to 80%. The company is working with TRIC, a United Nations partner, to create a digital artwork of Du Fu Thatched Cottage in Chengdu. Through a partnership with Amap, immersive 3D environments are now live in Amap's "Flying Street View" feature, which reaches nearly one billion monthly active users. Kosmo supports two business models: integrated hardware and software solutions, and services producing high-quality real-world spatial data at scale. Hesai's lidar deliveries have doubled for nine consecutive years, and the company holds a leading global market position in passenger-vehicle ADAS and robotaxis. The company is expanding its robotic actuation business, with modules powering Sharpa's Wave dexterous hand, which delivers three times the power density of comparable products in a package 37% smaller. The hand has been incorporated into NVIDIA's Isaac GR00T humanoid robot reference design and adopted by research institutions including UC Berkeley and Stanford. Cumulative deliveries of Hesai's actuation modules exceeded 10,000 units by the end of Q2 2026, and shipments are expected to reach six figures in 2027.

Disagree with this article?