VNET Group (NASDAQ: VNET) rallies as CATL stake and 517 MW order book reset AI thesis

VNET Group shares climbed 8.72 percent on Wednesday to $10.76, extending a multi-week rally that began when CATL-linked investors agreed to buy roughly 38 percent of the China-based data center operator in a deal valued near $1 billion. The move continued one day after the company reported first quarter results in which wholesale revenue overtook retail for the first time, validating its pivot toward hyperscale AI infrastructure. VNET is a carrier and cloud neutral internet data center services provider headquartered in Beijing, and the next major catalyst is the expected fourth quarter 2026 closing of the CATL transaction, which would reshape ownership and potentially unlock new customer pipelines.

What does VNET Group actually do and why is wholesale data center capacity in China repricing now?

VNET Group operates internet data centers across China on a carrier and cloud neutral basis, meaning it sells colocation, hosting, cloud and VPN services without favoring any single telecom or cloud customer. The company runs two business lines, retail colocation for smaller enterprise customers and wholesale capacity for hyperscalers and large internet platforms. Under what management calls a Hyperscale 2.0 framework, VNET has been redirecting capital toward larger, pre-committed builds rather than smaller retail deployments.

The competitive backdrop has shifted materially in the last twelve months. China’s policy support for artificial intelligence infrastructure, combined with utility and power quota constraints in core regions like the Greater Beijing area, has created a supply squeeze that benefits operators with existing land, power and project pipelines. Effective new supply remains limited, which is why VNET secured roughly 517 megawatts of new wholesale orders year to date 2026, including a 400 MW order and a 110 MW order from a single internet customer at separate data centers in the Greater Beijing area.

The retail investor angle is that VNET is being repositioned from a slow growing colocation provider into a leveraged play on China’s AI infrastructure cycle. The thesis depends on execution at the wholesale level, where contract sizes are larger but customer concentration risk is also higher. A single customer cancelling or delaying a 400 MW build would materially alter the trajectory, which is the kind of detail forum participants on Weibo, Xueqiu and English-language China stock channels are now scrutinizing.

Why did CATL-linked investors agree to buy nearly 38 percent of VNET?

On May 13, PJ Millennium entities affiliated with Contemporary Amperex Technology, the world’s largest electric vehicle battery maker known as CATL, agreed to acquire up to about 650.4 million Class A shares from existing holders at $1.4486 per share, or $8.6914 per American depositary share. The sellers are beneficially owned by Shandong Hi-Speed Holdings Group, the Hong Kong-listed infrastructure conglomerate. The transaction would give the CATL-linked buyers approximately 38.1 percent of VNET’s outstanding shares after closing.

Crucially, this is a secondary share purchase, not a primary capital raise. The cash flows to Shandong Hi-Speed, not into VNET’s treasury, which means the company itself does not directly receive new funding from the deal. What VNET gets is a strategic anchor shareholder with deep expertise in energy systems and a powerful customer rolodex. CATL’s core business is lithium-ion batteries for EVs and energy storage, and the read-across to AI data centers is energy management, where power density and grid efficiency are increasingly the binding constraints on hyperscale builds. The deal also includes investor rights and voting agreements that keep founder Josh Sheng Chen, who serves as interim CEO, in operational control.

The execution risk is twofold. First, the deal still requires Shandong Hi-Speed shareholder approval and is expected to close in the fourth quarter of 2026, leaving a six-month window in which conditions could change. Second, the strategic value of CATL as a shareholder depends on whether the relationship translates into actual customer wins or capacity commitments. The market is pricing the deal as a vote of confidence from a major industrial player, but the operational benefits will take quarters to verify.

How did VNET’s Q1 2026 print confirm or complicate the AI data center thesis?

VNET reported first quarter 2026 net revenue of RMB 2.69 billion, up 19.8 percent year over year, with wholesale revenue climbing 58.1 percent to overtake retail revenue for the first time in the company’s history. The shift in the revenue mix is the most operationally significant fact in the print. For years VNET was primarily a retail colocation business, and the crossover marks a structural change in what the company is and how investors should value it.

The headline numbers were softer. The company reported a net loss of $323.2 million, or $1.20 per share, on revenue of $390.1 million in USD terms, against analyst expectations of breakeven to slight loss. The loss was widened by tax expenses and the cost burden of building out new wholesale capacity ahead of revenue recognition. Management noted that the roughly 500 MW of new orders will be delivered in batches over 2026 through 2028, with the first batch expected to begin contributing in the second half of 2026.

The implication for retail investors is a classic build-ahead-of-revenue setup. VNET is spending on land, power and equipment now for orders that monetize one to three years out. This pattern compresses near-term margins and inflates capital expenditure, but if the precommitment rate management referenced on the call holds, the back half of 2026 and 2027 should show step-function revenue expansion. The risk is timing slippage, where construction delays or power quota issues push deliveries to the right.

What does the Q1 listing of two REIT projects mean for VNET’s capital structure?

In March 2026, VNET completed the listing of two real estate investment trust projects, which CFO Peter Zhihua Zhang described on the earnings call as establishing a scalable capital recycling model. The REIT mechanism allows VNET to monetize stabilized, cash-generating data center assets by selling them into a listed vehicle while retaining operational control, then redeploying the proceeds into new project development.

This matters because data center construction is capital-intensive, and VNET carries significant debt and has been burning cash. By recycling mature assets through the REIT structure, the company can fund new wholesale capacity without continually tapping equity or debt markets at the parent level. It also creates a benchmark valuation for its operating assets, which is useful both for negotiation with customers and for marking the implied value of unstabilized projects on the balance sheet.

The caveat for shareholders is that REIT monetization works only if the underlying asset cash flows are stable and the listed market for these vehicles remains liquid. China’s REIT framework is relatively new, and any pullback in investor appetite for infrastructure REITs would slow the capital recycling cadence. The strategy is sound in theory but unproven at scale through a full cycle.

How is VNET positioned versus GDS Holdings, Equinix and Digital Realty in the AI capacity race?

Within the China-listed data center landscape, VNET’s nearest comparable is GDS Holdings, also a wholesale-focused operator with significant hyperscaler exposure. Globally, Equinix and Digital Realty dominate the colocation and interconnection segments, but their China presence is limited by regulatory restrictions on foreign-owned data infrastructure. That leaves the China AI infrastructure opportunity to domestic operators, with VNET and GDS as the two scaled NASDAQ-listed proxies for the trade.

VNET’s differentiation rests on its dual-core strategy, retaining a substantial retail colocation franchise while building wholesale capacity for AI workloads. The retail business provides recurring revenue and customer diversification, while the wholesale build out captures the upside from concentrated AI capital expenditure. CATL’s incoming 38 percent stake also creates a differentiation that GDS does not have, namely a strategic shareholder with adjacent industrial capability in power and energy management.

The risk to the comparative thesis is that AI workload demand in China consolidates around two or three customers, primarily the major cloud platforms and large model builders. If a single hyperscaler controls the bulk of order flow, operators become highly dependent on that customer’s capex cycle, which makes earnings lumpy and reduces pricing power. The 400 MW order from one internet customer already signals significant concentration, and shareholders will want to see future order announcements come from different counterparties.

Why are retail investors on Reddit, Stocktwits and Chinese forums watching VNET ahead of the CATL closing?

VNET has become one of the more debated China tech ADRs on retail forums in 2026, sitting alongside names like NIO and PDD as a directional bet on China policy support and consumer cyclical recovery. The CATL stake purchase triggered a 25 percent single-day rally on May 13 when shares closed at $11.28 on volume more than seven times the three-month average. That kind of volume spike typically draws sustained retail attention because it indicates institutional repositioning, which retail traders often try to front-run on subsequent news flow.

Forum discussion has focused on three threads. First, whether the CATL relationship will translate into actual customer wins from CATL’s industrial network or its energy storage customers. Second, whether the company’s heavy debt load and negative free cash flow can be managed through the REIT recycling strategy without further dilution. Third, the timing and magnitude of the second-half 2026 revenue ramp as the first batch of new orders begins delivering capacity.

The execution risk for retail investors here is positioning fatigue. With the CATL deal not closing until Q4 2026 and the order ramp playing out over two to three years, the catalysts are spread thinly. Any quarter in which wholesale revenue growth disappoints, or in which a power quota issue delays a delivery batch, could trigger a sharp pullback even if the long-term thesis remains intact.

Key takeaways: What should investors watch as VNET moves toward the CATL closing and the order ramp?

• VNET rose 8.72 percent to $10.76 on May 27, extending a rally driven by the CATL-linked stake purchase and one day after a Q1 print in which wholesale revenue overtook retail for the first time.
• PJ Millennium entities linked to CATL agreed to acquire approximately 38.1 percent of VNET at $8.6914 per ADS from Shandong Hi-Speed Holdings, with closing expected in the fourth quarter of 2026.
• The cash from the CATL deal flows to Shandong Hi-Speed, not into VNET’s balance sheet, but the strategic shareholder is expected to provide energy expertise and potentially customer access.
• VNET secured 517 megawatts of new wholesale orders year to date 2026, including a 400 MW and a 110 MW order from a single internet customer in the Greater Beijing area, with deliveries staggered across 2026 to 2028.
• Q1 2026 net revenue of RMB 2.69 billion rose 19.8 percent year over year, with wholesale revenue up 58.1 percent, but the company posted a $323.2 million net loss widened by tax expenses and build-ahead costs.
• The March 2026 listing of two REIT projects establishes a capital recycling mechanism that could reduce VNET’s reliance on equity issuance to fund future wholesale capacity.
• The key risks are customer concentration in the wholesale book, execution timing on the multi-year delivery ramp, and the binary close of the CATL transaction itself, which still requires Shandong Hi-Speed shareholder approval.