Alibaba’s Profit Crashed 99% But Its AI Hiring Budget Doubled — How Foreign Companies Can Compete in China’s Salary Arms Race
On May 14, 2026, Alibaba reported its Q1 results: revenue up 3% to ¥243.4 billion, but non-GAAP net income collapsed from ¥29.8 billion to just ¥86 million — a 99.7% decline. That same day, Tencent posted revenue growth of 9% to ¥196.5 billion — but its operating profit would have been 8 percentage points higher without the drag from new AI products and services. JD.com fared worst: net profit halved, down 53% year-on-year.
Behind these numbers lies a counterintuitive dynamic. Tencent’s R&D spending jumped from ¥18.9 billion to ¥22.5 billion — nearly all of the ¥3.6 billion increment went to AI. Alibaba’s free cash flow turned negative for the first time. The money went to cloud and AI infrastructure, Qwen app acquisition costs, and Taobao’s instant retail subsidies.
China’s internet giants are trapped in an AI version of the prisoner’s dilemma. Skip the investment, and you’ll be irrelevant in 18 months. Make the investment, and your quarterly earnings take a beating. But for the talent market, this spending war has produced one unmistakable result: AI compensation has detached from every traditional salary benchmark.
AI Talent Salaries Haven’t Just Risen — They’ve Left the Old Grid Entirely
Recruiters at Sun Tzu China recently worked with a European industrial conglomerate to find an AI algorithm lead. The budget: ¥1.5 to 2 million per year. The candidate withdrew midway through the process — he accepted an offer from a major Chinese tech company: ¥1.8 million in cash plus equity, leading a team of 50.
This is not an isolated data point. Over the past twelve months, median compensation for top-tier AI talent in China has climbed 35 to 50 percent. A master’s graduate with three years of experience, a Big Tech internship, and one or two top-conference papers can now command ¥50,000 to ¥80,000 per month. In 2023, the same profile was worth ¥20,000 to ¥30,000.
The driver is a demand-supply mismatch that has no precedent in China’s tech labor market. AI-related job postings in China surged 12-fold between 2024 and 2026, with average monthly salaries running about 25 percent above the tech industry median. The supply side? Engineers who can genuinely train and optimize large language models — probably fewer than 50,000 nationwide.
The result is a compensation environment where traditional pay bands no longer apply. A role that would have been classified as P8 (senior) in 2023’s salary grid might now be filled by someone at a P6 or P7 skill level — but paid at P8+ levels. Because if you don’t pay it, ByteDance or Moonshot AI will.
China’s Tech Giants Are Trading Profits for AI Talent — and They Know It
Back to those earnings reports. The numbers tell a story that goes beyond quarterly volatility.
Alibaba’s non-GAAP net income fell by a factor of 300x year-on-year. Yet the company’s earnings call was unambiguous: “AI infrastructure investment is our highest priority for the next three fiscal years.” Tencent’s R&D spend increased 19%, and its marketing expenses surged on Yuanbao (its consumer AI app) user acquisition — including a reported several-hundred-million-yuan Lunar New Year giveaway campaign. JD.com’s AI investments are smaller in absolute terms, but its overseas instant retail push (Joybuy launched in six European countries in March) and AI-powered customer service systems are doubling spend year-on-year.
What does this mean structurally? China’s top tech companies are deliberately trading profit margins for AI talent density. They know this depresses stock prices — Tencent’s Hong Kong shares have corrected 24% from their 52-week high. Their calculus: the AI infrastructure race has an 18- to 24-month window, and losing that window means losing the next decade.
For foreign companies, this creates an unprecedented compensation dilemma. For the past decade, MNCs in China operated with a default salary anchor: 10 to 15 percent below local internet giants, offset by work-life balance, global exposure, and long-term career development. That anchor has lost its holding power in 2026.
Recruiters at Sun Tzu China found in a recent internal analysis that for AI and advanced algorithm roles, the gap between foreign company offers and local tech giant offers has widened to 30 to 50 percent. The “WLB premium” simply cannot bridge a half-million-yuan annual salary difference.
Three Strategic Responses for Foreign HR Leaders
This is not a call to match ByteDance dollar for dollar — no foreign company can win an AI talent bidding war with China’s domestic giants on cash alone.
But recalibrating the salary anchor does not mean chasing the top of the market. It means three things.
First, redefine your benchmark peer group. Most foreign companies still benchmark against other MNCs in China. That reference set is increasingly irrelevant. The smarter approach: dual-track benchmarking. Keep the traditional peer group for management and operations roles, but for AI and technical talent, use local tech company compensation bands as the ceiling reference — and plan accordingly.
Second, move from “salary” to “total rewards” with quantified precision. Chinese tech giants pay aggressively in cash, but foreign companies have structural advantages in long-term incentives, international rotation opportunities, and cross-border project exposure. The winning strategy: hold the cash gap to within 20 percent, then close the remainder with deferred equity, overseas training programs, and clearly defined cross-regional promotion pathways. The key is quantification — not “we offer global opportunities” but “you will spend six months at our Munich HQ within the next 18 months.”
Third, redesign AI role compensation bands entirely. Using traditional grade-level pay bands for AI positions is a self-elimination strategy. For core AI roles — LLM training, inference optimization, multimodal research — implement a separate approval track that bypasses standard HR grading. One-off, case-by-case compensation setting is not a sign of poor process; it is the only rational response to a market where supply is two orders of magnitude below demand.
The Hidden Advantage Foreign Companies Don’t Talk About
Here is a statistic worth noting. China produced more AI research papers than any other country in 2025, but its AI talent churn rate also hit an all-time high — over 35 percent annually. The same companies paying top dollar for AI engineers are losing them at an accelerating rate.
Why? Brutal work schedules, frequent research priority changes driven by executive whims, and chaotic internal management. One former AI Lab researcher at a major Chinese tech firm described his typical quarter: “Three direction changes, endless slide decks explaining to management what we do, zero time to actually build anything.”
Foreign companies have a structural but persistently undervalued advantage here: research direction stability, mature management processes, and an environment where engineers spend more time on technology than on organizational navigation. For a meaningful segment of AI talent — not everyone, but enough to be strategically important — this is worth more than an extra ¥200,000 to ¥300,000 per year.
The challenge is getting paid enough attention for the candidate to hear that story.
Three Actions for Q3 2026
First, run an AI compensation stress test. Pick your three most critical AI roles — algorithm engineer, AI product manager, data scientist — and reprice them against the May 2026 market. If any of your current employees are paid below 80 percent of market, treat them as immediate retention risks.
Second, establish an anti-poaching salary buffer. Set aside a quarterly reserve for emergency adjustments on key AI talent. Do not wait for the employee to come to you with a competing offer — by then, the departure probability is already above 70 percent. In Sun Tzu China’s case files, most retention counter-offers fail for one reason: lateness.
Third, quantify career development as a compensation equivalent. A foreign company employee can complete two international rotations in three years. A local tech company employee can get promoted once in three years and earn ¥300,000 more. Do not assume candidates intuitively understand the value of the foreign company path. Lay it out as a timeline with specific milestones, target geographies, and measurable outcomes — not a vague statement of “growth opportunities.”
——————————————————————————————
China’s tech giants will not stop burning cash on AI talent in 2026. The more their profits compress, the more aggressively they will spend on people — because for them, talent is not a cost line. It is the only ammunition in a war they cannot afford to lose.
Foreign companies do not need to join the arms race. But they do need to acknowledge that it exists — and adapt their compensation architecture accordingly.
An open question worth tracking through 2027: as AI transitions from “emerging field” to “table-stakes infrastructure” at every Chinese tech company, how long will the AI talent premium persist? The answer will determine the shape of every foreign company’s China hiring budget for the next three years.
Comments are closed