miHoYo invested in a loom company.

Not virtual weaving in a game — a real-world consumer smart loom that actually weaves scarves, hats, doll clothes, and pet accessories. The founder, Hu Wenxin, is an engineer who previously worked at DJI and Meituan. He founded CLAWLAB (Langzhua Intelligence) in late 2022 and emerged from three years of stealth for the first public reveal. The company has completed multiple rounds of over 100 million yuan in financing: Pre-A2 led by Yuanjing Capital with Shunwei Capital oversubscribing; Pre-A3 led by miHoYo, with Yuanjing and Shunwei significantly oversubscribing again; earlier investors include Sequoia.

Why Is a Smart Loom a Good Business?

Hu Wenxin didn't pull the consumer loom idea out of thin air.

During product definition, the team systematically mapped four major textile techniques — tufting, crochet, sewing, and weaving. Tufting, crochet, and sewing all add to or join existing fabric; they can't independently generate complete textiles. Only weaving can take yarn as raw material and produce finished garments, scarves, accessories, and plush toys in one shot, in a highly self-contained process — perfect for a desktop device where you "put in raw material, get a finished product."

The market size is striking. On Xiaohongshu, "weaving girl" topics have nearly 900 million views; crochet-related topics have over 3 billion total traffic. On TikTok, the "#crochet" tag has over 200 billion views. Core weaving enthusiasts globally number in the tens of millions, with a broader potential audience exceeding 100 million. But this market has long lacked modern products — home knitting machines briefly flourished in the 1990s (brands like Silver Reed, Brother) but were purely mechanical, fully manual, and saw essentially no systematic innovation in the three decades since.

CLAWLAB launched its first automatic tufting gun in 2024 for small-scale overseas validation, generating nearly 100 million yuan in revenue over two years. The real flagship product is the upcoming consumer weaving Station platform — a new form of physical computing terminal that requires no professional pattern-making; users can generate weaving patterns through simple drawing or photos, describe design intent in natural language, and a built-in textile AI Agent automatically parses styles, extracts features, and generates weavable plans, completing the full loop from "user intent" to "physical finished product." What takes a traditional pattern-maker days gets compressed to minutes in the Station.

AI and Robotics Sink into the Oldest Industry

Building a consumer-grade smart loom is far harder than outsiders imagine.

Weaving is a highly timing-dependent dynamic process — different pattern colors, different needles, interlocking rows; any tension or positioning error at any point can ruin the entire fabric. The mechanical action combinations for stockinette stitch and cable stitch are completely different; the tension control curves for chunky wool and fine cotton are completely different; the contour forming logic for a garment and a hat are completely different. No open-source algorithms to reference, no ready-made datasets to train on, no mature supply chain to reuse — the team spent nearly three years decomposing weaving techniques level by level into programmable control algorithms, transferring AI hardware capabilities like robotic control, motion planning, and computer graphics into product R&D.

Software and AI are CLAWLAB's deepest moat. The team developed its own textile AI Agent, a domain-knowledge-enhanced vertical model with built-in accumulated pattern-making algorithms and compilation capabilities. Users upload images or describe requirements in natural language, and the AI generates directly weavable design plans — the logic is identical to AI coding tools understanding requirements to generate code, except the output isn't code but fabric you can wear.

Why is miHoYo investing in looms? This seemingly cross-border choice actually has logic: miHoYo has long been interested in the intersection of AI + creation tools + consumer products. The weaving Station is essentially a "physical content creation platform" — users create designs, machines produce physical goods, communities share and exchange, small businesses customize and monetize. The ecosystem logic shares DNA with UGC gaming platforms. Sequoia and Shunwei and other top-tier firms repeatedly oversubscribing indicates they see the potential in this category.

CLAWLAB's user evolution path is clear: first wave is existing hardcore weaving enthusiasts (solving pain points), second wave is small-business operators (doll clothes, pet apparel, holiday customization, efficiency gains), eventually covering ordinary consumers (home micro flexible supply chain — want it, have it). This mirrors the path 3D printing took, but weaving's end products have more warmth — the comfort of a scarf, the companionship of a hat, the healing quality of a plush toy; these emotional values need no education.

While every founder is fighting over LLMs, AI coding tools, and Agents, Hu Wenxin chose the road less traveled: using AI and robotics to transform a 3,000-year-old traditional industry. It's a reminder: AI startups don't have to build LLMs. Bringing modern technology to forgotten traditional industries is an equally massive opportunity.

See you tomorrow.