Cover image 丨 Graphic by Qianbidao

AI companies are dangling hefty salaries to poach top talent, a trend that's also lining the pockets of headhunters.

Recently, Tech Planet reported that an embodied AI company was on the hunt for an AI algorithm expert, offering a total annual compensation package of 10.5 million yuan. With a headhunter fee rate of 26%, this translated to a service fee of 2.73 million yuan.

"I've never dealt with such a high-stakes placement," Yang Hongbo, co-founder of headhunting firm INR, shared with Qianbidao. Excluding future equity appreciation, the highest-value order he's actually handled was an overall offer exceeding 7 million yuan.

At a 25% fee rate, a single deal could bring in 1.75 million yuan for the company.

Yang Hongbo finds himself practically witnessing history on a daily basis.

Take this June, for instance, when a fresh PhD graduate from the University of Science and Technology of China was gearing up to join a company with a 620,000 yuan offer. Yang saw potential for a better fit and recommended him for another position, which immediately upped the offer to around 850,000 yuan.

Then, just three days before the scheduled start date, a third company entered the fray.

With an offer of 1.03 million yuan.

The candidate was poached once again.

"I poached someone, only to be poached myself in the end," Yang lamented.

In a matter of weeks, the PhD's technical skills hadn't magically improved, but the market's valuation of him had soared from 620,000 yuan to 1.03 million yuan.

- 01 - How Much Can a Headhunter Earn from a Single Deal?

"High-value deals do exist, but a 2.73 million yuan commission is extremely rare in the headhunting industry and not very representative," Yang Hongbo clarified to Qianbidao.

Based on his actual market experience, most headhunting orders generate service fees hovering around 100,000 yuan. Orders exceeding 300,000 yuan are already considered significant.

For some senior algorithm talents in the Agent field, annual incomes of 2-3 million yuan are not uncommon, with typical headhunter fees ranging from 20%-25%.

However, there's a catch when it comes to actual settlements.

Many AI talents receive "3 million yuan annual packages" that may include bonuses, stocks, and options. Yet, a large number of headhunting orders are still calculated based on fixed cash salaries.

For example, if a candidate's monthly base salary is 100,000 yuan, even with additional year-end bonuses and substantial equity, the headhunter fee might only be calculated as "100,000 yuan × 12 months × fee rate."

"The vast majority of orders we handle still follow this logic," Yang explained.

Therefore, the 2.73 million yuan is the headhunting firm's service fee, not the amount a single headhunter pockets.

In another report on employees of a related headhunting group, it was confirmed that the order existed, and it was estimated that after company and team divisions, the individual's actual income would be around 400,000-500,000 yuan. With typical industry fee rates of 20%-25%, traditional headhunters commonly achieve annual performances of 700,000-1 million yuan, with individual annual incomes mostly falling between 200,000-400,000 yuan.

Of course, the prices for top-tier talents have indeed skyrocketed.

In April this year, UBTECH publicly recruited an "Embodied AI Chief Scientist" with an annual salary starting at 15 million yuan and reaching up to 124 million yuan, consisting of cash, benefits, and equity. This role is responsible for defining the company's technical roadmap for humanoid robots and embodied AI.

But such a position is nearly equivalent to being the top technical executive of the entire company.

A case handled by Yang Hongbo better illustrates how the market prices ordinary high-end talents.

A technical lead from a major company originally had an annual income of around 3.6 million yuan. Later, a much smaller AI application company poached him.

The cash component remained largely unchanged at just under 2 million yuan annually.

The company's real bet was on equity, granting approximately 4.8 million yuan worth of shares at once. Based on the overall offer, it approached 7 million yuan.

The willingness of small companies to raise salaries to this level reflects a transformation occurring in the AI industry.

Yang Hongbo's most significant observation this year is that "AI application roles are beginning to surpass foundational model roles, which is different from last year. In the second half of this year, we'll see a much calmer market and talent pool, with a greater emphasis on practical application deployment and the choice of application tracks."

When companies are willing to pay a premium, it usually means the person is closer to products and revenue.

Headhunter commissions rise accordingly.

But this money isn't easy to earn.

Candidates backing out at the last minute, original companies offering raises to retain talent, competitors poaching, or even clients suddenly canceling newly assembled teams can all nullify months of tracking an order.

In the headhunting circle, this is called order jumping: while the windfall drives up prices, it also elevates risks.

- 02 - Poached Again Three Days Before Starting

A noticeable change in this year's AI talent market is that many people's prices are being recalculated based on their roles.

"In June this year, I encountered a fresh PhD graduate from the University of Science and Technology of China," Yang recounted.

With top-tier journal and conference publications, his main research was in graphics.

Following a traditional career path, the most obvious jobs for him were in game and graphics development. Upon graduation, a company offered him 620,000 yuan.

He was prepared to accept it.

Later, Yang contacted this candidate and realized his underlying technical skills could be applied to another type of R&D role, so he recommended him to another company.

After learning the original offer was 620,000 yuan, the new company immediately increased it by over 200,000 yuan.

To around 850,000 yuan.

The candidate felt the new role was a better fit and offered higher compensation, so he decided to switch.

Only three days remained before his official start date.

Suddenly, a third company emerged.

Offering a role in AI Infra with a salary of 1.03 million yuan.

Ultimately, he joined the third company.

"So I poached someone, only to be poached myself in the end," Yang chuckled.

The interesting aspect of this story is that the candidate hadn't suddenly published more papers or gained three extra years of work experience in those few weeks.

What changed was that his abilities were placed in different markets.

Graphics R&D: 620,000 yuan.

More foundational R&D role: 850,000 yuan.

Entering the then-scarcer AI Infra field: 1.03 million yuan.

Yang discovered that many fresh graduates with master's or doctoral degrees face issues where their skills aren't fully monetized.

"They know what they've studied, but not what else they can do. This is where the value of headhunters lies—we help candidates understand their market value and positioning," Yang emphasized.

This is precisely where the most intense competition lies in the current AI talent market.

Many companies aren't just vying for standard talents in standard roles but for individuals with transferable foundational skills.

This scarcity is also evident in the external market.

During this year's graduation season, Tsinghua University's Computer Science Department NLP Lab announced the career paths of 14 graduates, with four joining ByteDance and one joining DeepSeek. Liu Zhiyuan, a professor at Tsinghua's Computer Science Department, revealed that PhD graduates specializing in large models from the lab could earn annual salaries exceeding 6 million yuan, with master's graduates also potentially reaching the million-yuan level.

With few talents and many companies, offers naturally pile up.

Yang has seen the most extreme case where a candidate received nearly 20 offers in a month.

That individual had already decided to leave his original company, had little work to do, and spent his days in a phone booth attending interviews.

One or two interviews a day for half a month.

He ultimately accumulated nearly 20 offers.

However, Yang cautions that having many offers doesn't necessarily reflect the individual's scarcity but could indicate "changes in supply and demand" in the business sector.

If someone happens to align with the year's hottest directions, such as AI applications, Agents, or Workflow, they might be approached by over a dozen companies even with a monthly salary of only 30,000-50,000 yuan.

When it comes to top-tier talents, companies compete on more fronts.

Salary, equity, and position are just the most visible aspects.

Candidates also ask about: the progress of business commercialization, the company's cash flow, its investor background, the level of authority granted by the boss for this role, and the decision-makers' patience with the endeavor.

Sometimes, HR can't persuade them anymore, and the boss has to step in personally.

On the surface, AI talent poaching seems to be about raising salaries, but ultimately, it's about competing for a small group of individuals who can determine a project's success or failure.

- 03 - The Stronger AI Gets, the More High-End Headhunters Earn

The headhunting industry is also being impacted by AI, but there's no need for excessive anxiety. Yang Hongbo also believes that while AI will automate a large amount of repetitive work, the true value of high-end headhunters still lies in connections and career insights.

Nowadays, AI can already search resumes, match candidates to roles, and screen applicants. Specialized Agent products have also emerged in the recruitment industry.

However, when a company encounters a candidate holding a dozen offers, convincing them to join is the hardest part.

Corporate HR might contact dozens or even hundreds of candidates a day, sending company introductions and continuing conversations only with those interested.

Headhunting work is more like sales.

When a truly suitable candidate emerges, it requires repeated communication: why they want to leave, why they don't, what they're worried about, their salary expectations, and whether they prioritize business or position.

Yang believes there's another often underestimated value here—career planning.

Opportunities and risks at Company A; higher pay at Company B but with the possibility of the business disappearing in two years; this candidate currently works in graphics, but are there other companies in the market willing to pay more for his skills?

All these factors influence the final deal.

The hotter the AI industry, the more critical this capability becomes.

In Tech Planet's report, a headhunter mentioned that some companies, eager to recruit quickly, would assign a single role to numerous headhunters simultaneously, leading to situations where "hundreds of headhunters compete for one role." For the same candidate, whoever recommends them to the company's talent pool first might secure the order.

At this point, simply finding a phone number offers little competitive advantage; the real challenge is poaching the candidate and ensuring they stay.

Here's the irony: while recruitment is one of the industries most easily enhanced by AI, AI headhunters are actually doing better business this year.

The headhunters who ultimately stay are those providing candidates with high-quality judgment, negotiating on their behalf, and completing human-to-human connections.

This article does not constitute any investment advice.