SenseTime Open-Sources 1 AI Model to Compete With Qwen

SenseTime has released and open-sourced its SenseNova U1 model, a strategic move to capture market share in a landscape increasingly defined by low-cost, high-performance models from competitors like Alibaba and DeepSeek. The release of the unified model for understanding and generation, built on an architecture developed with Singapore's Nanyang Technological University, signals a shift toward building a wide developer ecosystem rather than focusing on short-term licensing fees.

"The path to monetization is more measured than the capital markets priced in," said Joseph Alagna, Founding Partner at Buttonwood Funds, in a recent commentary on the AI sector. "That's not a reason to exit, it's a reason to be selective." This sentiment captures the strategy behind open-sourcing: a long-term bet on market expansion and platform adoption over immediate revenue, a path that requires significant investment and patience.

The SenseNova U1 model is built on the NEO-unify architecture, which integrates multimodal understanding, reasoning, and generation within a single framework. While SenseTime has not disclosed specific performance benchmarks, it enters a fiercely competitive open-source arena. Alibaba’s Qwen model family surpassed 1 billion cumulative downloads by March 2026, capturing over 50% of all global open-source model downloads. Meanwhile, Chinese rival DeepSeek recently cut prices on its V4-Pro model to be nearly 35 times cheaper than comparable models from OpenAI.

For investors in SenseTime (00020.HK), this move is a clear pivot toward a platform-based strategy. By making its model freely available, the company aims to replicate the success of Alibaba's Qwen in building a vast community of developers who use its technology as a foundational layer. The bet is that this will drive future demand for SenseTime's broader portfolio of AI products and services, even as the cost of individual model usage collapses toward zero.

The Open-Source Arms Race

SenseTime's decision was not made in a vacuum. It comes as the open-source AI landscape is being rapidly reshaped by Chinese technology firms. While much Western attention has focused on the drama at OpenAI, including its slowing growth and legal battles, Alibaba's Qwen has quietly become the dominant force in open-weight AI. According to an MIT Technology Review analysis, Chinese open-weight models, led by Qwen, surpassed the US in global download share for the first time in 2025.

This dominance was achieved through a deliberate strategy of releasing over 100 open-weight models under the permissive Apache 2.0 license, covering every tier from mobile devices to large-scale data centers. This has made Qwen the default starting point for a new generation of AI development globally, with governments like Singapore and Malaysia choosing it for their sovereign AI initiatives. SenseTime's open-sourcing of SenseNova U1 is a direct response to this new reality, an attempt to build its own gravitational pull in a market where developer mindshare is the ultimate prize.

Jevons Paradox and the AI Economy

The collapsing cost of AI tokens, exemplified by DeepSeek's aggressive price cuts, is often misinterpreted as a bearish signal for the industry. However, some of the industry's biggest players see it differently. Microsoft CEO Satya Nadella and former Intel CEO Pat Gelsinger have both invoked the Jevons Paradox, an 1865 economic theory stating that as technology becomes more efficient, its total consumption increases. As the cost of using AI models falls, the number of viable applications expands exponentially.

This is the economic principle underpinning SenseTime's strategy. Cheaper, open-source models are not a threat to the AI economy; they are the substrate that will allow AI to be embedded into everything. By contributing to this trend, SenseTime is positioning itself to capture value not from selling access to a single model, but from the explosion of new use cases that its technology will enable. The challenge will be converting that broad adoption into revenue for its enterprise-grade offerings and infrastructure.

Success for SenseTime will depend on its ability to foster a vibrant developer community around SenseNova, a difficult task given the commanding lead established by Alibaba's Qwen. The move pressures short-term revenue but is a necessary strategic play to remain a relevant competitor in the next phase of AI. For investors, the focus shifts from model-to-model comparisons to tracking the growth and engagement of the SenseNova ecosystem.

This article is for informational purposes only and does not constitute investment advice.