On May 9, JD.com officially launched global recruitment for its 2026 TGT (Top Young Tech Talent) program. This is the second consecutive year JD has run the program, and compared with last year it adds an intern track, extending recruitment to students graduating in 2027 and beyond.

In 2026, as large-model adoption enters deeper waters, AI talent supply lags far behind the pace of demand growth. JD’s escalation is a step forward in its technical talent pipeline, and it also reflects how fierce the scramble for core AI talent has become across China’s tech industry.

What TGT Is, and How This Year Differs from Last Year

TGT is JD’s dedicated recruitment program for undergraduate, master’s, and doctoral students worldwide — current students, recent graduates, and technical talent within two years of graduation — positioned around “top young tech talent.” Compared with regular campus hiring, TGT offers significantly more in compensation, resources, and development structure.

When it first launched in 2025, TGT only targeted recent graduates and technical talent within two years of graduation. The key change in 2026 is the new TGT intern track:

The intern track means JD wants to engage with top students’ development paths earlier. For high-potential candidates who haven’t graduated yet, JD offers a channel into its technical system before the formal job search. According to i 黑马 (iheima) reporting, the goal of this change is to “lock in high-potential talent early.”

Uncapped Pay — and What Comes with It

“Uncapped compensation” is TGT’s most shareable tagline. But for top technical talent, salary is just the ticket in; the real draw is research resources and growth space. JD pairs the pay with three supporting conditions:

A dedicated, ultra-large-scale compute cluster. Compute is one of the core bottlenecks in AI research. JD provides TGT selects with dedicated ultra-large-scale compute resources for large-model training and inference experiments. Per 21 经济网 (21jingji) reporting, this resource is quite competitive given the industry-wide compute crunch.

Dedicated young-talent housing. Addressing housing reduces the cost-of-living pressure on young researchers.

A three-mentor system. Every TGT member gets a “technical mentor + business mentor + growth mentor.” This mechanism splits academic guidance, business application, and career development into three dimensions, avoiding the crude “one mentor does everything” model. Per the official TGT QA on Nowcoder, the three-mentor system is the core development mechanism distinguishing TGT from JD’s regular campus hiring.

Eight Technical Tracks, All Pointing at AI

This round of TGT recruitment covers eight technical tracks, spanning nearly every key layer of today’s AI stack:

1. Multimodal large models and applications
2. Machine learning
3. Search, recommendation, and advertising
4. Spatial and embodied intelligence
5. High-performance and cloud computing
6. Big data
7. AI infrastructure (AI Infra)
8. Security

Among these, multimodal large models, spatial and embodied intelligence, and AI Infra are new or highlighted tracks this year. These map closely onto JD’s own business: unmanned delivery and warehouse robotics in logistics correspond to embodied intelligence, multimodal understanding on the e-commerce platform corresponds to large-model applications, and the large-scale online transaction systems create a natural need for AI infrastructure.

Per Tencent News reporting, JD also added “enterprise-grade AI model training” as a key direction, meaning TGT isn’t purely academic — researchers are also expected to be able to engineer models into production.

Talent Bar: Papers, Patents, Competition Awards

TGT’s screening criteria favor candidates with strong academic and competition backgrounds. Per official information republished by NetEase, selectees need at least some of the following:

• High-quality, high-impact papers in authoritative journals or top conferences
• Technical patents in relevant fields
• Major awards in well-known technical competitions
• Deep insight into a specific technical domain, willing to break from conventional frameworks

These criteria echo ByteDance’s Top Seed program and Alibaba’s “Ali Star” (阿里星) program — all use academic output and competition results as “hard currency” to quickly identify top talent.

The Big-Tech AI Talent War, in Panorama

JD’s TGT is far from alone. From 2025 to 2026, major Chinese tech companies rolled out dedicated AI talent programs with escalating benefits:

Per 21jingji’s analysis, these programs share an arms race across three dimensions: compensation ceilings keep rising, compute resources are becoming a standard perk, and mentorship or development systems are the differentiating selling point.

As pay and compute converge, the real differentiator shifts to who can offer more real-world industry data and more complex business scenarios. JD’s edge lies in the massive amounts of real data its logistics and e-commerce systems generate every day, plus engineering needs across real-time recommendation, warehouse scheduling, and unmanned delivery. For AI researchers, these scenarios mean the possibility of “research driven by business.”

Public Reaction: Skepticism Behind the High Pay

TGT’s launch also sparked discussion. Reading across comment sections on multiple platforms, the criticism clusters around three points:

The graduation window is too narrow. TGT campus hires must have graduated within two years, which some voices criticize as veiled age discrimination. Commenters note the two-year limit excludes many experienced senior candidates with proven track records.

The sustainability of the high pay is questionable. The “uncapped” phrasing invites speculation about actual salary levels. Some argue that “hire high, lay off quickly” can’t be ruled out.

Is the AI infrastructure sufficient? JD isn’t an AI company in the traditional sense; some commenters are skeptical about its compute reserves and technical depth.

These reactions point to a deeper issue: as “uncapped pay” becomes uniform messaging across the big tech firms, its marginal appeal is diminishing. Which company a candidate ultimately picks comes down to the authenticity of the research scenarios and the depth of the technical teams.

Strategic Intent: JD’s Position in the AI Race

As of the end of 2024, JD’s cumulative R&D investment approached 140 billion RMB. The TGT program is JD translating R&D spending into talent pipeline — a concrete action.

From a broader view, TGT can be read as part of JD’s technology-brand building. In the large-model era, the flow of top technical talent often decides a company’s standing in the AI race. With the combination of “uncapped pay + dedicated compute + real business scenarios,” JD is trying to win more top talent attention against Alibaba, ByteDance, and Meituan.

Per analysis cited by Tencent News, TGT is essentially an upgraded version of JD’s earlier PhD management-trainee program. But the “genius program” narrative, plus the expanded intern track, gives it significantly more reach in both buzz and talent coverage.

How to Apply

TGT applications have two official channels:

• PC: visit JD’s campus recruitment site campus.jd.com; the “Top Young Tech Talent Program” section on the homepage supports direct online applications
• Mobile: follow the “京东招聘” (JD Recruitment) WeChat official account, go to the “校园招聘” (Campus Recruitment) item in the bottom menu to submit your resume

Campus hires and interns are both open year-round with no fixed deadline. TGT interns don’t need to take the online assessment; campus hires complete the online assessment before moving to interviews.

References

• i 黑马 — JD recruits “AI geniuses” with uncapped salary and dedicated compute
• 21 经济网 — The big-tech AI talent war heats up
• Tencent News — JD recruits “AI geniuses” with uncapped salary and dedicated compute
• NetEase — JD launches the “Top Young Tech Talent Program” globally: uncapped compensation
• Nowcoder — JD TGT Top Young Tech Talent Program QA
• Longbridge — JD.com Launches the “Top Young Tech Genius Program”