SenseTime’s Galaxy Project targets domestic AI chip scale-up

SenseTime has launched the Galaxy Project, teaming with almost 20 companions to scale domestic AI chip infrastructure in China.

In a keynote titled ‘Intelligent Transformation and Symbiosis,’ Yang Fan – the corporate’s co-founder and president of its Large Device Business Group – laid out what SenseTime describes as a closed loop connecting chip-level expertise, ecosystem partnerships, and business deployment for domestically-produced AI computing energy.

Alongside the Galaxy Project, SenseTime signed an area computing settlement with satellite tv for pc producer Guoxing Aerospace and struck a analysis partnership with 5 establishments – together with the Shanghai Artificial Intelligence Laboratory – aimed toward scientific computing purposes.

Yang framed the timing round three converging tendencies: token demand climbing throughout enterprise deployments, industrial AI adoption catching up with consumer-facing use instances, and domestic chip commercialisation reaching some extent the place clever computing centres constructed on Chinese silicon could be stood up at tempo.

However, whether or not that window is as open as SenseTime claims relies upon closely on numbers the corporate has not had independently verified.

Token throughput figures include a big asterisk

SenseTime says its large-scale system platform now processes a median of two.42 trillion tokens each day, and the corporate initiatives that determine will climb 25-fold to 10 trillion tokens per day by the fourth quarter of 2026. That’s a forecast, not a measured outcome, and enterprise patrons evaluating SenseTime’s infrastructure ought to deal with it as such till quarterly figures begin touchdown.

The cost-effectiveness claims connected to that development are equally self-reported. SenseTime says its heterogeneous hybrid inference expertise delivers an 85–152 % improve in Model FLOPs Utilisation on mainstream domestic chips, alongside inference cost-effectiveness the corporate places at 1.25x that of Nvidia’s H-series components.

Compared with domestic homogeneous inference setups, SenseTime claims a 2.5x improve in token output at equal price, a leap it says pushes optimised hybrid inference clusters previous what the business beforehand considered the minimal profitability threshold for domestic computing energy.

None of those figures include third-party benchmarking, and the hole between a vendor’s optimised take a look at cluster and a buyer’s manufacturing atmosphere – with its uneven information pipelines and delayed firmware updates – tends to be the place such numbers soften.

Adaptability claims and the multi-chip drawback

Domestic AI chips have traditionally struggled with a fragmented software program stack: fashions educated for one structure usually require rework to run on one other. SenseTime says it has constructed a full-stack adaptation layer spanning fashions, frameworks, operators, toolchains, and {hardware} to handle that, with the intention of letting clients migrate workloads throughout domestic chip distributors with out intensive rewrites.

The firm factors to 2 utilized examples. In an AI4S long-sequence protein prediction workload, SenseTime says fused operator optimisation minimize total prediction time by an element of three. In AIGC video technology, it claims a 93 % multi-card parallel acceleration ratio for domestic chips working DiT fashions, alongside what it describes as zero-cost migration for mainstream AI improvement instruments.

These are the sorts of figures that learn effectively in a sandbox take a look at and matter way more as soon as they’re stress-tested towards actual buyer pipelines working combined {hardware} generations.

Energy metrics get a brand new benchmark title

SenseTime launched a metric it calls Tokens Per Watt, positioned as a substitute yardstick for measuring AI information centre effectivity, alongside a Computing-Power Collaboration Agent that handles useful resource scheduling, electrical energy worth prediction, and vitality storage optimisation throughout what the corporate describes as an eight-level information system with 5 resolution chains.

Combining compute, electrical energy pricing, and automatic scheduling, SenseTime claims an 80 % improve in token output per unit of electrical energy price, common energy costs 10 % under comparable regional information centres, and 96 % accuracy in computing load prediction.

These are claims price watching over the subsequent a number of quarters moderately than accepting at face worth. Electricity worth arbitrage and cargo forecasting accuracy are inclined to carry out in another way as soon as a system runs by a full seasonal cycle with real demand volatility, moderately than the circumstances underneath which a vendor sometimes runs its pilot.

Impressive accomplice roster spans chipmakers to element suppliers

The Galaxy Project’s said ecosystem consists of domestic chip distributors Cambricon, Muxi, Hygon, Huawei Ascend, Moore Threads, Sunrise, and Biren Technology, element accomplice Xizhi Technology, and infrastructure companies together with Silicon Motion, Qujing Technology, Zhongke Jiahe, Qingcheng Jizhi, Sophon Information, and Jiliu Technology.

SenseTime says the plan covers building of 1 “token manufacturing unit,” 5 computing clusters at what it calls “10,000-calorie” scale, joint work throughout ten expertise instructions, and help for 200 AI startups.

“Domestic manufacturing shouldn’t be merely about changing particular person chips, however moderately a collaborative effort throughout all the chain of China’s innovation capabilities, from chips and elements to infrastructure and utility situations,” Yang stated.

Space, optical, and quantum computing bets look additional out

Beyond near-term infrastructure, SenseTime outlined work on optical computing for information centre effectivity, quantum computing purposes in AI optimisation, and an area computing partnership with Guoxing Aerospace to construct what the 2 firms name the SenseTime Space Computing Constellation.

SenseTime’s plan requires a primary satellite tv for pc launch in 2026, constructing towards 1000’s of computing satellites and computing capability within the tens of 1000’s of petabytes by 2030.

Yang argued the worth extends previous uncooked functionality, framing space-based computing as a option to prolong the attain of Chinese AI companies into weak-network environments akin to maritime operations and catastrophe response, and by extension to help China’s AI exports internationally.

That 2030 goal sits 5 years out, and satellite tv for pc computing deployments of this scale don’t have any precedent to measure the timeline towards.

Physical infrastructure spans Shanghai to Riyadh

On the bottom, SenseTime says its Shanghai facility runs the nation’s first information centre rated at what it calls “5A” clever computing degree, dealing with over 20 trillion tokens each day throughout greater than 20 industries. A Yancheng web site has launched with an preliminary 3,000 petaflops of capability targeted on vitality, manufacturing, and low-altitude financial system purposes.

In Hong Kong, SenseTime is constructing what it describes because the territory’s largest domestic clever computing centre, concentrating on 40,000 petaflops by 2030. The firm additionally plans what it calls China’s first abroad domestic computing cluster in Saudi Arabia, positioned as a full-stack domestic computing base for the Middle East.

On the analysis facet, SenseTime’s tie-up with the Shanghai AI Laboratory, Beijing Zhongguancun Academy, Shenzhen Hetao Academy, the Shanghai Algorithm Innovation Research Institute, and Shanghai Jiao Tong University’s AI college goals to construct a shared platform spanning compute, tooling, and mannequin functionality for all times sciences, supplies science, and manufacturing analysis. Yang referred to as AI for Science “a key lever for paradigm innovation in primary analysis,” tying the initiative to China’s broader “Artificial Intelligence+” coverage push.

SenseTime’s forecast of 10 trillion tokens per day by This autumn 2026 is the determine to trace towards regardless of the firm reviews when that quarter really closes.

See additionally: Kimi K3 open-weight model: China’s biggest AI is a bet on memory, not compute

Want to study extra about AI and large information from business leaders? Check out AI & Big Data Expo going down in Amsterdam, California, and London. The complete occasion is a part of TechEx and is co-located with different main expertise occasions together with the Cyber Security & Cloud Expo. Click here for extra info.

AI News is powered by TechForge Media. Explore different upcoming enterprise expertise occasions and webinars here.

The submit SenseTime’s Galaxy Project targets domestic AI chip scale-up appeared first on AI News.