NextFin News - Kuaishou Technology's second-quarter profit fell by roughly a third as the Chinese short-video group's aggressive push into artificial intelligence and rising payouts to content creators ate into margins faster than revenue grew. Net profit for the three months ended June 30 dropped 36% to RMB 3.15 billion, the company said in its August 19 results filing, while revenue edged up just 1.4% to RMB 35.54 billion — a top-line figure that landed essentially in line with analyst expectations but did little to offset the earnings shock.
The Numbers: A Deliberate Trade-Off
The headline figures show a company choosing growth over near-term profitability — and paying for that choice in real time:
- Net profit: RMB 3.15 billion, down 36% from RMB 4.9 billion a year earlier
- Adjusted net profit: RMB 3.9 billion, down 30% from RMB 5.6 billion, though marginally ahead of consensus estimates of about RMB 3.91 billion
- Total revenue: RMB 35.54 billion, up 1.4% year on year, essentially in line with the RMB 35.51 billion consensus
- Gross margin: 51.6%, down from 55.7% a year earlier — 410 basis points of margin erosion in a single quarter
- Operating profit: RMB 3.8 billion, down from RMB 5.3 billion
- Adjusted EBITDA: RMB 7.12 billion, down 7.7% from RMB 7.7 billion
Zooming out to the first half of 2026, the picture is consistent rather than anomalous. Revenue for the six months ended June 30 rose just 2.4% to RMB 69.3 billion, while adjusted net profit fell 28% to RMB 7.3 billion from RMB 10.2 billion. This is the second consecutive quarter of profit declines in the 26-36% range against revenue growth under 4% — a pattern, not a one-off.
The revenue split reveals where the pressure is coming from. Online marketing services — the advertising engine — grew just 4.4% to RMB 20.6 billion. Live streaming revenue, once the cash cow, shrank 13.5% to RMB 8.7 billion as the company continues to prune its creator ecosystem. Only "other services," which bundles e-commerce with the Kling AI video-generation business, showed real momentum, up 18.5% to RMB 6.2 billion.
Kling AI itself generated more than RMB 850 million in revenue during the quarter, up over 200% year on year. That sounds like the growth story investors want to hear. The problem: it is still a rounding error against a RMB 35.5 billion revenue base, while the spending required to build it is anything but small.
Research and development expenses jumped 26.5% quarter-on-quarter to RMB 4.58 billion — well above the RMB 4.07 billion consensus — with the company citing "increased investment in AI, including related training spend." Meanwhile, cost of revenue rose 4.5% sequentially to RMB 17.2 billion, driven, in the company's own words, by "revenue-sharing costs and related taxes increasing with revenue."
Two engines are burning cash at once: the AI arms race and the creator economy that funds it. And the overseas business, a hoped-for second leg of growth, swung back to a small operating loss of RMB 25 million after a modest profit a year earlier.
The Two-Engine Cost Squeeze
Kuaishou is being squeezed from both sides of its income statement, and the mechanism is worth separating because the two cost pressures behave very differently.
On the AI side, the spending is discretionary and strategic. Every yuan poured into model training, GPU clusters, and data centers is a bet that Kling AI can become a standalone growth pole — or at least a defensible moat around the core advertising business. This is capital the company chooses to deploy. It is also, critically, front-loaded: training spend hits the profit and loss statement now, while the revenue payoff, if it comes, arrives later and uncertainly.
On the creator side, the spending is structural and harder to switch off. Revenue-sharing costs — the payouts to the creators whose videos keep users on the platform — are the price of admission in a market where ByteDance's Douyin, WeChat Channels, and a host of smaller rivals are all bidding for the same attention. When Kuaishou says these costs rose "with revenue," it is describing a cost structure that scales up automatically: more activity on the platform means more money owed to creators. That is not a line item management can trim with a single decision. It is the operating leverage of the creator economy working in reverse.
The combination is what makes this quarter uncomfortable. A company can justify a temporary AI investment spike to shareholders. It can also explain away rising creator payouts as ecosystem building. But both accelerating simultaneously, while revenue growth stalls at 1.4%, is the textbook definition of an earnings-quality problem.
Growth at What Price? The Margin Trade-Off
The 410-basis-point gross margin decline is the clearest single read on what is happening. Gross margin compression of this magnitude does not come from one-off items or accounting noise. It comes from a business model where the cost of delivering each additional yuan of revenue is rising faster than the revenue itself.
Put differently: Kuaishou is buying growth at an increasing marginal cost. The RMB 850 million Kling AI revenue is growing at more than 200%, but the R&D and infrastructure required to produce it is growing faster still — R&D alone up 34.7% year on year. The advertising business, which should be the high-margin engine funding the transition, is growing at a single-digit pace.
There is a precedent investors should remember. In the first quarter, Kuaishou already posted a 26.3% decline in adjusted net profit, to RMB 3.37 billion, on revenue growth of just 3.4%. Two quarters in a row of profit declines in the 25-36% range, against revenue growth under 4%, is not a cyclical wobble. It is evidence of a deliberate strategic pivot — and of the market beginning to price in what that pivot costs.
This is not an isolated story. Across China's internet sector, the same tension is playing out: mature cash-generating businesses are being asked to fund AI build-outs whose payoffs are years away. Kuaishou's 2026 capital expenditure plan of roughly RMB 26 billion — about RMB 11 billion more than 2025 — sits in the same family as the spending commitments that rattled investors at other Hong Kong-listed tech names this year. The difference for Kuaishou is scale: with a revenue base an order of magnitude smaller than the sector's giants, each yuan of AI capex moves the margin needle more violently.
Kling AI: The Growth Engine That Costs More Than It Earns
Kling AI is, by any operating metric, the success story of this earnings release. Revenue up more than 200%. An annualized revenue run-rate approaching half a billion dollars as of March. Product launches including native 4K video generation and a turbo model aimed at cost efficiency. Recognition at the Cannes Creative Lions for AI-generated advertisements.
But the unit economics of the AI video business remain the question nobody in the earnings call wanted to answer directly. Video generation is among the most compute-intensive AI workloads in existence. Every second of generated video carries a real inference cost, and Kuaishou's own filing points to "training spend" as a primary driver of the R&D surge. The company has said Kling has reached gross-margin positive territory on inference — but that is inference only, not the full-stack cost of research, training runs, and the capital expenditure required to keep pace with ByteDance's Seedance and the rest of the global field.
The cost curve matters as much as the revenue curve here. Training runs are lumpy and upfront; inference costs scale with usage. If Kling's growth is driven by heavy discounting to win creators, the revenue line can rise while the contribution margin stays thin. Management's stated goal of lowering unit training and inference costs over the medium term is the hinge the whole thesis swings on — and it is a goal, not yet a demonstrated result.
This is where the second-order effect bites. The market is not just asking whether Kling can grow. It is asking whether Kling can grow without permanently depressing the parent company's margin structure. Kuaishou's answer so far has been to explore external financing for the AI unit — a raise reported at up to about $3 billion that would, if completed, put the heaviest capital expenditure onto outside investors' balance sheets rather than Kuaishou's own.
That is a rational response. But it is also an admission: the AI dream is too expensive to fund from the cash flows of a short-video app growing at 1.4%.
The Counter-Thesis — This Is a Plan Executing, Not a Breakdown
The strongest argument against reading this quarter as deterioration is that Kuaishou's management is doing exactly what it said it would do. The company has been explicit that 2026 capital expenditure would reach roughly RMB 26 billion, about RMB 11 billion above 2025 levels, with the incremental spend directed at Kling and other foundation models. Chief Financial Officer Jin Bing framed it at the first-quarter earnings call as using "today's precise investment to leverage future profit growth."
We will use today's precise investment to leverage future profit growth.
From this vantage point, the profit decline is not a surprise — it is the plan executing. Revenue-sharing costs rise because the platform is investing in creator quality over quantity, a deliberate ecosystem upgrade. R&D rises because AI is the assigned priority. The balance sheet can absorb it: RMB 121.3 billion in available funds, RMB 1.97 billion spent on share buybacks in the first half, and no interim dividend declared, preserving cash for the build-out.
That buyback is worth reading carefully. A company that is genuinely worried about a cash crunch does not spend nearly RMB 2 billion repurchasing its own shares in six months. Management is signaling that the earnings pressure is a known, budgeted transition cost — not a liquidity event.
This counter-thesis has real force. Kuaishou is not running out of money, and it is not losing its way operationally. Daily active users grew 1.0% to 412.5 million and monthly active users surged 11.5% to 797.3 million — user engagement, the foundation of everything, is still expanding.
But the counter-thesis rests on one assumption: that the AI investment converts into durable, high-margin revenue before investor patience runs out. That is the same assumption every capital-intensive technology transition requires, and it is the assumption that turned ByteDance's own reported profit decline earlier this year into a market-moving event — a reminder that even the deepest-pocketed players are not immune when AI spending outruns revenue.
What Would Prove the Thesis Wrong
The bear case — that Kuaishou is sacrificing profitable, mature businesses to fund an AI venture whose economics are not yet proven — would be falsified by one specific signal: Kling AI reaching sustained, full-stack profitability, covering training plus inference plus allocated capital expenditure, while core advertising revenue re-accelerates to double-digit growth.
Conversely, the bull case breaks if revenue-sharing costs and AI training spend continue to compound at 20% or more annually while total revenue stays in the low single digits for another two quarters. At that point, the margin compression stops being a transition cost and becomes the business model.
Outlook: Three Time Horizons
Short term (next one to two quarters): expect continued pressure. Management has already signaled a cautious stance on the second half of 2026, citing a complex and challenging macro environment. With the heavy capital expenditure weighted toward the first half of the year, the second half should see some moderation in the spending growth rate — but profit comparisons will remain difficult against a year-ago period when margins were still intact.
Medium term (six to eighteen months): everything hinges on the Kling AI financing and the margin trajectory. If the external raise closes at the rumored valuation, Kuaishou's own profit and loss statement gets relief and the stock could see a re-rating on sum-of-the-parts logic. If it stalls, the market will force the choice management has so far avoided: slow the AI build-out or accept structurally lower margins.
Long term (structural): the creator-economy cost structure is the real regime change here, not AI. Once a platform's cost base becomes mechanically tied to revenue-sharing with creators, operating leverage never fully returns to its pre-creator-economy level. The question for long-term holders is not whether Kuaishou can cut its way back to 55% gross margins — it almost certainly cannot — but whether Kling can become large enough that the group's margin profile begins to look like an AI company's rather than a short-video app's.
Base case: revenue grows at mid-single digits through 2026, adjusted profit remains under pressure at roughly RMB 15-18 billion for the full year, and the Kling financing closes, capping the downside.
Upside case: Kling revenue doubles again in the second half, advertising re-accelerates on AI-driven targeting, and the external raise values the unit above $20 billion — a sum-of-the-parts re-rating follows.
Downside case: creator costs and AI spend keep compounding while revenue stagnates; the financing round is delayed or downsized; multiple compression extends into 2027.
Kuaishou is not losing the AI race — it is paying the entry fee, and the bill is coming due faster than the revenue it was supposed to buy.
Explore more exclusive insights at nextfin.ai.
