CATL's first-half numbers still look like an EV battery story at first glance. Look closer and you see the better story: energy storage revenue jumped 87.54% just as AI data centers are becoming a power infrastructure problem.
Most people tracking CATL still think of it as an EV battery company. That framing is old. When CATL reported its first-half 2026 results on July 24, Reuters put the main figures plainly: net profit rose 42% year-on-year to 43.3 billion yuan, or about $6.39 billion, while revenue climbed 54.8% to 276.9 billion yuan. Strong enough. But the number that tells you where the company is moving sits inside energy storage, where revenue rose 87.54% to 53.26 billion yuan, according to CnEVPost's reading of the half-year report.
The comparison with CATL's original business is the point. Its power battery unit generated 192.12 billion yuan in revenue in the first half, up 46.02%. That is still the larger business by a wide margin, but it isn't the fastest one anymore. Energy storage is now growing almost twice as fast as the division that made CATL the world's largest battery maker.
At the same time, CATL announced a plan to buy back 20 billion to 40 billion yuan of its A-shares and cancel them. That is not a casual footnote. Companies use buybacks that large when management wants to send a message about value, confidence and capital discipline. You don't need to romanticize it. The cash is real.
Margins are the other useful detail. CnEVPost reported CATL's overall gross margin at 23.93%, down 1.09 percentage points from a year earlier, while energy storage gross margin was 23.96%. That small decline matters because China's EV battery market is still dealing with brutal pricing pressure. If CATL can keep roughly 24% gross margin while energy storage nearly doubles, the market has to stop treating the company as only a supplier to carmakers.
AI needs batteries, not just chips
The AI infrastructure buildout has a power problem that doesn't get enough attention. GPU clusters running at scale need more than raw electricity. They need stable power delivery, backup capacity and grids that can handle sudden, heavy loads. Battery storage sits in that gap between intermittent supply and the constant draw that AI compute demands.
CATL has spent 2026 moving toward that gap with money, not slogans. In April, it invested about 4.1 billion yuan, roughly $600 million, in Hangzhou Zhongheng Technology Investment for a 49% stake, according to ESS News. Zhongheng Electric, in which that vehicle is a major shareholder, makes high-voltage direct-current power systems and has served data center customers including Alibaba and Tencent. That is the kind of dull infrastructure detail investors should love. It connects directly to where AI facilities hurt.
Then came VNET. A May 13 company filing said PJ Millennium I and PJ Millennium II, whose parent partnership is tied to a non-controlled and non-consolidated CATL affiliate, agreed to buy up to 650.4 million VNET Class A shares from existing holders. The price was $1.4486 per ordinary share, or $8.6914 per ADS, and Bloomberg put the potential deal value at about $942 million. If it closes in the fourth quarter of 2026, the buyers would hold up to about 38.1% of the Chinese data center operator.
That's a serious position. VNET said in its first-quarter results that it expects 2026 revenue of 11.5 billion to 11.8 billion yuan and capital expenditure of 10 billion to 12 billion yuan. You don't attach yourself to that kind of data center capex unless you believe power systems and energy storage will become part of the AI stack, not an afterthought bolted on later.
DeepSeek adds the louder name. Reuters reported in early June that DeepSeek was set to raise about 50 billion yuan, or $7.4 billion, from investors including Tencent and CATL. Pandaily later reported that CATL and its affiliates invested about 5 billion yuan in the round. Frankly, the strategic logic is obvious: DeepSeek needs computing clusters, computing clusters need reliable power, and CATL wants to be closer to both the batteries and the customers consuming them.
The sodium-ion signal
The product roadmap is beginning to match the investment trail. Energy-Storage.news reported that CATL showed a dedicated sodium-ion cell for grid-scale storage at ESIE 2026 in Beijing, with more than 300 ampere-hours of capacity, 97% efficiency and more than 15,000 cycles. The same report said the cell can cover two-hour to eight-hour storage use cases, including AI data center scenarios, with commercial rollout planned for 2026.
CATL's own announcements back up the wider push. On June 22 in Munich, the company unveiled its TENER Sodium Energy Storage System and said deliveries in China would begin in September 2026, with cumulative shipments expected to reach 1 GWh by the end of the year and global deliveries starting in June 2027. In April, CATL and HyperStrong also signed a three-year, 60 GWh sodium-ion energy storage agreement. Those are not laboratory talking points. They are delivery schedules and purchase commitments.
Kevin Tang, CATL's director of energy storage systems for Europe, told Reuters in June that the company expects energy storage to account for half of global sales by 2030, up from about 25% now. That target sounded aggressive a month ago. After the first-half results, it looks less like ambition and more like a business already changing shape.
The EV price war in China is real. Competitors are discounting. CATL's gross margin is still sitting near 24%, and the adjacent energy storage business is growing at 87.54%. That is the detail to hold onto. Saturation in one market looks different when the next market is already arriving, and CATL has spent the year buying its way toward the data centers that will need power before they need another press release.
Also read: Anduril is in talks to raise money at a $100 billion valuation and the deal structure is unlike anything private markets have seen • Samsung put USDC into 241 million Galaxy phones and Circle got the distribution win crypto has been chasing for years • The EU found TikTok guilty of failing to protect children and every consumer app founder in Europe should take note
