Li Luwei spent nine years shipping Ningbo outdoor furniture to European customers. A 2025 conversation in Morocco about DeepSeek told her the next cross border play was AI distribution — and she moved to London to prove it.
In early 2025, Li Luwei was on a routine business trip in Morocco when she noticed something that did not require a market research report to confirm. On café Wi-Fi and in trade-floor conversation, the talk kept drifting to DeepSeek, the Chinese AI lab whose low-cost, openly released model weights had begun rippling through emerging markets faster than the Western press was registering. 'Openly released' here means the model's parameters are published for anyone to download and run, which is unusual for a frontier AI lab and is the reason a Casablanca café can be running a Chinese model on a local server. For nine years Li had run Import Market Sales, a B2B2C cross-border e-commerce platform shipping Ningbo outdoor furniture to French and UK marketplaces and direct-to-consumer customers. The Moroccan conversations told her the next cross-border opportunity was no longer in the container.
By 2026 she had left Paris for London to join ByteDance, where her remit is AI large-model global partnerships and ecosystem development, the team responsible for stitching foundation models, the large general-purpose AI systems that other companies build on, into outside products and developer communities. The operator who had spent a decade mapping European last-mile delivery was now mapping which AI builders, distributors and integrators ByteDance should be working with. The move, in Li's own account, is the second of her 'two outings': a deliberate retelling of the cross-border script (36Kr founder account, translated).
The setup Li is now walking away from was substantial by the standards of small-team European e-commerce. Import Market Sales, the platform she founded in 2017, linked the Ningbo outdoor-furniture industrial cluster to French and UK marketplaces, accumulating roughly €35 million in revenue over nine years on no external funding. Its owned brand, Avril Paris, ran the direct-to-consumer side. The company signed partnerships with more than 20 European e-commerce platforms and served over 80,000 customers, with a core team of seven. The credibility markers were real: Station F, the Paris startup campus founded by Xavier Niel, admitted IMS in 2017, and the French Ministry of Economy, through its French Tech program, recognized the company as an 'innovative enterprise' in 2020 (36Kr).
Li argues the transferable skill is the operator muscle that 2010s Chinese cross-border founders built and that 2025's AI distribution problem seems to demand. Shipping a Ningbo sofa to a Lyon balcony forces the same kind of decisions that putting a Chinese foundation model in front of a Casablanca developer does: how to slot a foreign-built product into a non-Chinese channel, how to read regulators and partners across time zones, and how to keep unit economics honest at small average order sizes. The Morocco conversation was the trigger; her argument is that the trigger applies to an entire cohort of operators who shipped physical goods in the 2010s and are now being courted for AI work.
The 36Kr piece is a single self-narration by a single founder. Revenue, partnership and customer counts, as well as the Station F and French Tech recognitions, are reported in Li's voice and have not been independently audited in the materials available. The broader claim, that a generation of 2010s Chinese cross-border operators is rotating into AI labs and AI-distribution roles, is plausible but unverified beyond this one legible case. Reporting it as a pattern requires either a second named operator making a comparable move or an industry dataset on the cross-border-to-AI pipeline; neither is in hand.
What is supported by the public record is a specific, dated action: a Paris-based Chinese cross-border founder with nine years of European e-commerce operating history, French Tech recognition, and no outside funding, joining ByteDance's London AI partnerships team in 2026 to work on foundation-model ecosystem development. Whether Li's move represents a sector-wide rotation from goods to AI is her argument, not this article's.
Whether a second or third operator from the 2010s Chinese cross-border wave announces a comparable move into an AI lab, AI-distribution partnership, or AI-enabled cross-border stack in the next 12 months is the data point that would turn this from a career-change story into a structural one.