DeepCtrls, a Nanjing company focused on AI control for energy and infrastructure, has reported a Series B+ investment in the several-hundred-million-yuan range, led by CATL. TMTPost reported the funding on September 8 and said Fosun Capital, Taiping Innovation Investment, and GF Xinde participated, with existing shareholders including Source Code Capital increasing their stakes.
The round brings attention to an overlooked part of the physical-AI economy. Much of the public discussion centers on models, robots, and chips. But the facilities that make AI possible also need to manage electricity, cooling, equipment behavior, and operating costs. DeepCtrls is trying to apply machine learning and physical-system knowledge to that layer, serving data centers, semiconductor sites, and new-energy facilities.
The company’s financing does not demonstrate that it has solved those problems across the industry. The size of the round was not disclosed more precisely than multi-hundred-million yuan, and the company’s performance claims need to be treated cautiously. Still, CATL’s role as lead investor offers a concrete example of a major energy and battery company taking an interest in systems that connect AI demand with energy-management infrastructure.
An AI layer for energy and control systems
TMTPost describes DeepCtrls as working on energy efficiency, intelligent control, and infrastructure optimization. Those areas are increasingly relevant as data centers add denser AI workloads. Training and inference systems can consume large amounts of power, generate heat, and require careful coordination among servers, cooling equipment, and electricity supply.
DeepCtrls calls its core system PhyAI. The company describes it as a way to combine machine learning with physical mechanisms in order to optimize real-world control systems. That description should be understood as product positioning rather than independently verified performance. In an industrial setting, a reliable control platform needs to do more than make a useful prediction. It must operate safely, explain or validate its recommendations, and work with existing equipment and operating procedures.
The company says its focus includes liquid cooling and coordination between computing and power resources. Those are plausible targets for AI-enabled optimization. Cooling can represent a large component of data-center operations, and matching computing workloads with energy availability may become more valuable as AI infrastructure expands. But the economics will vary by site, hardware configuration, power market, and operational constraints.
China’s policymakers have already emphasized the connection between AI capacity and energy. EastFrontier’s examination of the 15th Five-Year Plan’s AI-energy synergy showed why electricity, data centers, and industrial systems are becoming a shared strategic concern. DeepCtrls is positioned directly in that overlap.
CATL’s investment signals an industrial interest
CATL is best known as a battery manufacturer, but its investment in DeepCtrls fits a broader industrial logic. Energy systems are becoming more software intensive, while AI facilities are becoming more energy intensive. A company that can make infrastructure more responsive to conditions in real time could have value across both areas.
TMTPost said the funding will support ongoing research, product scaling, and market expansion. It also reported that the B+ round followed a Series B financing earlier in 2026. That sequence shows the company has attracted repeated investor attention, though it does not tell outsiders how much revenue it earns or how widely its systems are deployed.
A China Daily Chinese-language page also reported the round and named Aramco Ventures among participants. That page is explicitly labeled promotional content, so the article should not rely on it as an independent confirmation. The core facts used here are limited to TMTPost’s account: CATL led the round, named investors participated, and existing shareholders added capital.
The careful distinction matters. Funding announcements often mix verifiable transaction details with broad claims about market leadership, customer reach, and technology performance. For a company operating in complex physical systems, those claims need to be tested over time by measurable outcomes such as energy savings, reliability, customer retention, and safety performance.
Physical AI needs dependable infrastructure
DeepCtrls’ story shows why physical AI is not limited to humanoid robots. A control system that improves how cooling equipment, power assets, or industrial operations respond to changing conditions is also operating in the physical world. Its success depends on connecting data to action under real constraints.
That is a different market from general-purpose chatbots. The models may need to work with sensor streams, engineering rules, and equipment that cannot simply be rebooted when an experiment fails. Customers will demand confidence that automated decisions do not disrupt production or create safety risks. In those conditions, the value of AI is tied to trustworthy integration rather than a dramatic model demonstration.
China has seen growing investment in the surrounding infrastructure for embodied and industrial AI. ScaleForce’s funding for embodied-AI data infrastructure highlighted another part of the stack, focused on the data needed to train physical systems. DeepCtrls addresses the operational side, where AI must coordinate with energy and computing assets already in use.
The Series B+ round gives DeepCtrls more capital to test that position. Its future will depend on whether it can translate an attractive industrial narrative into repeatable deployments. CATL’s investment may bring strategic credibility and possible industry connections, but it does not eliminate the technical and commercial hurdles.
As AI computing expands, the companies that manage its physical footprint may become more important. DeepCtrls is betting that control systems, not only chips and models, will be a central part of that future. The new funding gives the company a stronger mandate to prove it.