Nike scores 82/100 in Chinese AI visibility while Lululemon — despite 47% China revenue growth — scores just 44. We break down why legacy content depth beats commercial momentum in China's AI search era.

Sportswear GEO in China: Why Nike and Adidas Still Win the AI Visibility Race — Even as They Lose Market Share

Key finding: Nike and Adidas each score above 78/100 in average Chinese AI brand visibility despite losing ground to domestic competitors. Lululemon, with 47% China revenue growth last year, scores just 44/100 — a visibility gap that risks capping future growth as AI-driven discovery becomes the norm.

China's sportswear market is set to hit $55 billion in 2026, making it one of the most contested consumer arenas in the world. The headlines have been dramatic: Nike saw Greater China sales fall double digits, Adidas scrambled to reclaim relevance post-Xinjiang controversy, and domestic giants Anta and Li-Ning have taken serious market share. Meanwhile, Lululemon has been on a tear — revenues in Mainland China rose 47% year-over-year in constant currency terms.

But here's what the sales figures don't tell you: in the Chinese AI search ecosystem, the old hierarchy still holds. The brands that built deep Chinese-language content ecosystems years ago continue to dominate how China's 900+ million AI search users encounter sportswear recommendations — regardless of what is happening to market share on the ground.

This is the sportswear GEO paradox: winning in stores no longer guarantees winning in AI answers.

How Chinese AI Models Score the Sportswear Category

According to hubGEO brand visibility data, which tracks how six major Chinese AI models (Doubao, Kimi, DeepSeek, Qwen, Wenxin, Hunyuan) respond to category queries and brand-specific searches, sportswear brands cluster into three distinct tiers.

Tier 1 — AI Dominant (75–100): Nike (~82), Adidas (~78)

Tier 2 — Domestic Challengers (55–74): Anta (~71), Li-Ning (~68), FILA (~62)

Tier 3 — Underrepresented (below 55): Lululemon (~44), New Balance (~51), Under Armour (~29)

The pattern reflects something consistent across all categories hubGEO tracks: Chinese AI models draw heavily from the depth and breadth of Chinese-language source material, not simply from global brand prestige. Nike's decades of Weibo campaigns, Tmall flagship stores, Douyin athlete collaborations, and localized product launches in Chinese have created a dense citation network across every major AI training corpus. Adidas, despite its well-documented controversies, maintained enough Chinese-language presence to remain in Tier 1.

Why Doubao's 260M Users Make This Urgent

Doubao (ByteDance) surged to 260 million monthly active users in Q1 2026 — a 300% increase from 2025 and now the clear leader in China's AI assistant market, ahead of Baidu ERNIE Bot at 220M MAU. For sportswear brands, this matters because Doubao's training universe is the ByteDance content stack: Douyin, Toutiao, and Xigua Video.

A brand without a strong Douyin presence is systematically underweighted in how Doubao forms opinions about category leaders. Lululemon's Douyin account, while growing, has significantly less depth of engagement and localized content than Nike's or Adidas's — and this structural gap shows up directly in Doubao's brand citations.

Run a query like "哪个运动品牌瑜伽裤最好" ("which sportswear brand has the best yoga pants") across Chinese AI models, and the results are revealing: Doubao reliably surfaces Lululemon, but often as a secondary mention after domestic alternatives, whereas Nike and Adidas get primary placements for broader athletic categories with supporting context drawn from video and editorial content.

The implication for Lululemon: even with genuine consumer enthusiasm and rapid sales growth, the brand's AI citation depth has not yet caught up to its commercial momentum.

Platform-by-Platform Breakdown for Sportswear

Doubao (260M MAU, ByteDance): Rewards brands with rich Douyin and Toutiao presence. Nike and Adidas are heavily favored. Anta and Li-Ning perform above their global status here because of deep native content. Lululemon visibility is moderate but growing.

Kimi (90M MAU, Moonshot AI): The long-context specialist is most likely to draw from in-depth editorial comparisons, gear reviews, and structured content like articles on 小红书 (Xiaohongshu). This platform is where Lululemon performs best in the sportswear category, because yoga and fitness lifestyle content on Xiaohongshu has begun building a meaningful citation trail for the brand. Score: ~53/100.

DeepSeek (rapidly growing, enterprise-leaning): Used heavily for structured research queries. When buyers ask "top international sportswear brands for enterprise procurement", DeepSeek references brands with clear commercial identity in China — consistent retail presence, verified Tmall flagship, media coverage. Nike dominates. Lululemon and Under Armour underperform significantly.

Qwen (Alibaba ecosystem): Naturally integrates Taobao and Tmall data. Brands with comprehensive Alibaba ecosystem presence — product depth, review density, brand story pages — score better. Adidas maintains a strong position here due to its Alibaba ecosystem integration. Anta, which sells heavily through Taobao, gets an unexpected citation boost.

Wenxin / ERNIE (Baidu, 220M MAU): The legacy enterprise leader. References Baidu Baike entries, Baidu Index data, and brand-specific news coverage indexed by Baidu. Nike's 百度百科 (Baidu Baike) entry is extensive; Under Armour's is minimal and rarely updated.

Hunyuan (Tencent, 150M MAU via Yuanbao): Integrates WeChat ecosystem signals. Brands with active WeChat service accounts, mini programs, and WeChat Pay integration score higher. Adidas' WeChat mini program is a category benchmark; Lululemon's WeChat presence is functional but lacks content depth.

The Domestic Brand Factor: Anta and Li-Ning Punching Up

One finding that surprises Western brand managers: Anta (~71/100) and Li-Ning (~68/100) consistently outperform Lululemon and Under Armour in Chinese AI citations — despite having far lower global brand recognition.

The reason is structural. Both brands produce content natively in Chinese for Chinese audiences, reference Chinese athletic culture, sponsor Chinese Olympic athletes, and operate within the same digital ecosystems that feed AI training corpora. Li-Ning's collaboration with Chinese fashion (the "国潮" wave) created an enormous body of editorial and social content that AI models frequently reference when discussing sportswear innovation.

For international brands, this creates an uncomfortable truth: domestic competitors have a native GEO advantage in Chinese AI that requires deliberate effort to overcome.

New Balance: The Viral Trap

New Balance presents an instructive cautionary tale. The brand achieved significant cultural virality in China — associated with vintage aesthetics and a certain "dad shoe" aesthetic that became fashionable. However, hubGEO data shows New Balance scoring inconsistently across Chinese AI models (~51/100 average), with its strongest performance on Kimi and Doubao and its weakest on Wenxin and Hunyuan.

The reason: viral social moments generate spike citations rather than sustained citation depth. AI models weigh recency and breadth of coverage — but they also weight structured, authoritative content. New Balance's Chinese-language brand content lacks the institutional depth of Nike or Adidas, meaning that outside of direct brand name queries, it is often absent from category recommendations.

Virality in China's social ecosystem is valuable but does not automatically translate into AI citation authority.

Takeaway for Brand Marketers

The sportswear data reveals a principle applicable across categories: AI visibility in China is a lagging indicator of content investment, not a current-state reflection of commercial performance.

Lululemon's 47% China revenue growth has not yet produced the content ecosystem depth needed for proportional AI citation authority. The risk is real: as Chinese consumers increasingly begin brand discovery via Doubao, Kimi, and DeepSeek — rather than direct search or WeChat ads — a brand's AI citation rate will increasingly predict its organic discovery rate.

For sportswear brands looking to close the GEO gap, three actions matter most:

Invest in Douyin content depth, not just reach. Doubao's 260M users make it the highest-impact platform for consumer sportswear categories. Long-form product reviews, athlete stories, and training content in Mandarin build the citation trails that AI models use.
Build Kimi-optimized long-form content on Xiaohongshu. For lifestyle and premium sport categories, Xiaohongshu editorial content gets disproportionate weight in Kimi citations. A structured "brand guide" format with comparison context performs better than campaign posts.
Audit Baidu Baike and Tmall brand pages. Wenxin and Qwen — which together serve 400M+ MAU — draw heavily from Baidu's indexed content and Alibaba's commerce ecosystem. An under-optimized Baidu Baike page is leaving AI citations on the table.

The brands that close the distance between commercial performance and AI citation authority fastest will be those that treat GEO not as a marketing project but as core infrastructure for China market success.

Track real-time brand visibility scores across Doubao, Kimi, DeepSeek, Qwen, Wenxin, and Hunyuan at hubGEO /brands. Sportswear category data is updated quarterly.