Kingsoft Cloud heads into its August 19 report with an unusual split screen: the operating story has arguably never looked better, yet the stock has fallen 12.3% since the last print while the S&P 500 rose 3.6%. That 16-point gap between fundamentals and share price is the central puzzle investors need to resolve this quarter.
Consensus calls for revenue of $447.8 million, up 36.6% year over year, building on last quarter's 37.2% growth in RMB terms. A GAAP loss of $0.12 per share is expected, narrower than last year's $0.21 loss but slightly wider than last quarter's $0.13. In other words, Wall Street is modeling continued top-line acceleration alongside a loss that isn't shrinking as fast as growth is expanding. That combination puts real emphasis on margins, not just revenue.
The last call was a genuine inflection point. AI gross billings crossed 50% of public cloud revenue for the first time, up from 39% a year earlier, and management described token/inference demand as supply-constrained rather than demand-constrained, with April revenue for its Starflow MaaS product running 53 times January levels. Adjusted EBITDA margin jumped to 27.6% from 16.2% a year prior, and capex guidance was effectively doubled to RMB15-20 billion for the year. Those are not incremental updates. They represent a company telling investors its growth ceiling just moved higher.
The catch, and the reason this report matters so much, is that adjusted gross margin cratered sequentially last quarter as AI server costs and upfront customer onboarding expenses bit into profitability. Management framed this as temporary, promising a return to "normal levels" in coming quarters. This report is the first real chance to test that promise. If gross margin stabilizes or improves even modestly while AI revenue keeps compounding, the narrative of profitable AI-driven growth gets validated. If margin pressure persists or worsens, investors will reasonably start asking whether the capex ramp is outrunning the company's ability to monetize it efficiently.
Supply chain constraints are the other swing factor. Management was explicit that backlog exceeds what it can currently deliver due to chip and component shortages, with upstream prices still rising. Any commentary on easing capacity, new supply agreements, or pricing pass-through will matter more than the headline revenue beat itself, since it speaks to whether growth is capped by demand or by hardware availability, a much more solvable problem if managed well.
Sentiment data suggests the market isn't dismissing the story. Earnings Whisper sentiment sits at +0.642, up from +0.430 last quarter, indicating expectations have grown more optimistic even as the stock price has slid. That divergence is worth sitting with. It suggests some investors see the AI ramp as durable while the broader market has been more skeptical or simply repricing risk around capex intensity and margin visibility.
Technically, the stock trades at $11.41, below its 200-day moving average of $12.61, and sits well off both its post-earnings high of $14.10 and its previous quarter's high of $18.52. The trading range has compressed significantly versus the prior quarter, and shares are closer to the low end of the current range than the high, meaning there's no breakout setup to lean on here. This looks more like a stock searching for confirmation than one anticipating it.
The real question for this report isn't whether AI revenue keeps growing, it almost certainly will. It's whether Kingsoft Cloud can show that scaling AI infrastructure doesn't require permanently sacrificing gross margin, and whether supply constraints are easing enough to let backlog convert into recognized revenue. Get those two things right, and the disconnect between operating momentum and stock price becomes an opportunity rather than a warning sign.