Few stocks have endured a sharper disconnect between what management says and what the market believes than GDS Holdings. Three months ago the Chinese data center operator doubled its full-year bookings target to 1 gigawatt, raised capital spending and lifted revenue and EBITDA guidance. Its tone was the most confident in at least four quarters. Since then the shares have lost about half their value, trailing the S&P 500 by roughly 45 percentage points. The third-quarter report, due before the open on November 25, has to show whether that bullish narrative still holds or whether investors were right to bail out.
On the numbers, Wall Street expects revenue of about $474.4 million. That is a 17% gain from a year earlier and a modest step up from the $455.1 million reported last quarter. Consensus earnings of $1.31 per share imply growth of more than 190% from $0.45 a year ago. A jump that large on mid-teens revenue growth suggests the bottom line may reflect items beyond core data center operations, so the per-share figure is probably a weaker guide than revenue and EBITDA. The more telling comparison is guidance. Management now expects 2026 revenue of $1.88 billion to $1.93 billion, yet the full-year consensus sits around $1.84 billion, below even the low end. Analysts appear unwilling to fully credit the raise. That leaves room for an upside surprise if the company simply delivers on its own range.
The heart of the story is bookings. GDS signed a record 470 megawatts in the first half, 260 of them in the second quarter, and said it had secured another 600 megawatts of reservations with more than 1 gigawatt expected by year-end. Reaching 1 gigawatt for the year implies roughly 530 megawatts in the back half. Third-quarter signings that keep pace with the second quarter would go a long way toward validating the target. A noticeable slowdown, or reservations that fail to convert into firm contracts, would feed the suspicion that the target was ambitious marketing. Investors should also look for more detail on the newer customer relationships with emerging AI players and on whether all three of the largest hyperscalers are still adding capacity. Rising concentration among those top customers is a risk worth tracking.
Near-term revenue was never supposed to be dramatic. Management guided to 235 megawatts of net move-ins for 2026, about the same as 2025, with only 90 megawatts in the second half. The real acceleration is not expected until the second half of 2027. Monthly service revenue per square meter is also expected to fall about 3%, pressured by cheaper new-market locations and the remaining repricing of legacy contracts. Those headwinds are known. What would challenge the thesis is a steeper pricing decline, a slip in move-in timing, or any sign that domestic GPU supply, which management said was catching up, is again holding customers back. Backlog growth from the 757 megawatts reported at mid-year would be the clearest offset.
The balance sheet matters too, given a higher capital spending target. Net leverage had improved to 4.7 times EBITDA, and confirming that trend while funding a bigger build-out would ease worries. Progress on the first post-IPO asset injection into the company's REIT vehicle, still under regulatory review, would show a recycling path for capital.
Sentiment has swung from modestly bullish before the last report to clearly bearish now. The stock, at $16.41, sits less than 6% above its post-earnings low and far below its 200-day average near $35. Expectations, in other words, are depressed. The question is whether third-quarter bookings and backlog confirm that GDS is building the multi-year growth pipeline it described. If they do, the gap between fundamentals and price looks hard to justify. If they don't, the selloff will look like a warning the market heeded early.