SenseTime to raise $415m via share placement

SenseTime Group, a Chinese AI software and hardware company, plans to raise HK$3.25 billion (US$415 million), through a Hong Kong share placement.

The funds will be used for AI model and cloud infrastructure spending, as well as general working capital.

The firm will sell 1.7 billion shares at HK$1.90 (US$0.24) each, an 8.6% discount on April 16, and the new shares will represent about 4.04% of its enlarged share capital.

SenseTime said 40% of the proceeds will go to its SenseCore infrastructure platform and AI cloud stack, with the rest for generative AI research, commercial applications, and working capital.

The placement does not need shareholder approval because it falls under a mandate approved at last June’s annual meeting, allowing up to 7.4 billion new shares, with 3.42 billion have been issued so far.

SenseTime shares fell 4.8% to HK$1.99 (US$0.25), taking their drop this year to 9.6%.

🔗 Source: South China Morning Post

🧠 Food for thought

Implications, context, and why it matters.

This capital raise follows operating gains and steadier cash flow

The share placement dilutes existing investors. It comes after SenseTime reported its first positive EBITDA (earnings before interest, taxes, depreciation, and amortization) and positive operating cash flow in the second half of 2025 1.
SenseTime plans to put the proceeds into faster execution, including scaling SenseCore, SenseTime’s AI cloud infrastructure platform, plus the related AI cloud stack 2.
The company also wants a higher “localization ratio” for large-scale computing infrastructure. The term means a larger share of systems built with domestic technology and suppliers, tied to this type of placement 3.

SenseTime funding speeds a homegrown AI stack that competes with Nvidia

The investment also matters for China’s domestic AI chip industry, which aims to compete with Nvidia, the leading US AI chip company 3.
SenseTime said SenseCore ran large-scale heterogeneous mixed training on a cluster of 5,000 domestic GPUs (graphics processing units). The approach trains AI systems on different chip types together, reached up to 80% computing power utilization, plus 95% of homogeneous training efficiency, which uses one chip type 4.
The deal also offers a precedent for US-sanctioned tech firms seeking local funding. SenseTime went ahead with its Hong Kong initial public offering (IPO) in 2021 after US sanctions delayed pricing 5.
The capital supports infrastructure buildout plus commercialization work that SenseTime says it needs for domestic chips and domestic models. This reinforces the move toward more self-sufficient technology stacks, meaning integrated sets of hardware and software, in China 3.