Goldman Sachs comments on the "Battle of China's AI Giants": Alibaba vs Tencent vs ByteDance
```
If 2024 was a “battle of model parameters,” then by late 2025, China’s AI battlefield will have evolved into a contest of capital efficiency, infrastructure hegemony, and traffic gateways.
On November 27, Goldman Sachs released a major research report on the development of China’s internet and artificial intelligence, analyzing the fierce “battle of tech giants” in China’s current AI industry and interpreting the different strategic choices of Alibaba, ByteDance, and Tencent:
- Alibaba chose the heaviest path, with capital expenditure soaring 80% year-on-year and a full-stack deployment, aiming to build a Google-like “full-stack” barrier with heavy assets to become the “full-stack overlord” of China’s AI market;
- ByteDance leverages its terrifying traffic advantage, with a daily consumption of 30 trillion tokens, making a huge impact at the application layer;
- Tencent has maintained its usual restraint, cutting capital expenditure while focusing on “seamlessly embedding” AI capabilities into its vast social and payment ecosystem.
At the same time, Goldman Sachs thinks that competition between US and Chinese AI is entering a new “dynamic alternation” normal: when US models reach new heights, Chinese models quickly iterate and catch up within the following 3-6 months.
Alibaba’s “full-stack advance,” Tencent’s “restraint,” ByteDance’s “traffic breakthrough”
1. Alibaba: A heavy “full-stack” bet matching Google
Alibaba is adopting a Google-like full-stack approach.
Goldman Sachs’ report highlighted Alibaba’s capital expenditure: Alibaba’s September quarterly capital expenditure surged 80% year-on-year to RMB 32 billion.
According to Goldman Sachs analysts, despite chip supply volatility, Alibaba is firmly executing its AI infrastructure expansion plan, trying to build a path similar to Google with “foundation model + multimodal capabilities” vertically integrated into “dedicated chips + full-stack services,” to consolidate its dominance in the enterprise market.
This “heavy asset” strategy is being redeemed in real money.
The report shows Alibaba Cloud’s external revenue grew 29% year-on-year in September, with AI-related revenue achieving triple-digit growth for nine consecutive quarters. Goldman Sachs predicts Alibaba Cloud’s revenue growth will accelerate to 38% in the December quarter, driven by strong AI training and inference demand.
On the consumer side, Alibaba is no longer low-key.
Its newly launched “Tongyi App” surpassed 10 million downloads in its first week; Ant Group’s “Lingguang” achieved 2 million downloads within six days; Quark app is also iterating rapidly with AI.
2. ByteDance’s “traffic breakthrough”
In the face of Alibaba’s offensive, ByteDance’s approach is marked by a distinctive “traffic gene”: using absolute dominance in consumer apps to feed its fundamental infrastructure.
The report reveals, ByteDance’s daily token usage has broken through 30 trillion. How big is that? It’s nearing global leader Google’s 43 trillion, far exceeding other Chinese peers—by comparison, Baidu and DeepSeek are at about 10 trillion daily tokens.
Goldman Sachs notes ByteDance’s “Doubao” ranks first in domestic AI app activity, and its overseas-focused educational app Gauth saw monthly transaction revenue rise 394% year-on-year, setting a historical high.
This huge inference demand from the application layer enables ByteDance to flank traditional cloud giants in the MaaS (Model-as-a-Service) segment.
Quoting IDC data, Goldman Sachs points out that in the “public cloud market share for large models”—a key future battleground—ByteDance’s Volcano Engine has secured 49.2% market share.
(3) Tencent: The “pragmatist’s ecosystem penetration”
In contrast to Alibaba and ByteDance’s aggressive moves, Tencent has maintained its customary “restraint” in the AI arena.
Goldman Sachs notes Tencent’s quarterly capital expenditure declined year-on-year, with its annual budget target also lowered.
Besides chip supply volatility, the report says this reflects Tencent’s “ecosystem penetration” strategy, focusing on seamlessly “stitching” AI into its large WeChat ecosystem.
The report notes Tencent has integrated its AI assistant “Yuanbao” into WeChat Pay, and leveraged AI copywriting tools and menu recognition features to directly boost operational efficiency for small and medium businesses. While this approach lacks eye-catching capex figures, it boasts highly reliable practical implementation.
US-China Model Race: 3-6 Month Dynamic Catch-up Cycle
Zooming out globally, Goldman Sachs also compared US and China AI technology dynamics.
Goldman Sachs analyst Ronald Keung’s team points out that despite concerns about diminishing marginal returns from the AI Scaling Law, Google’s recent launches of Gemini 3 Pro and the image generation model Nano Banana Pro once again demonstrate the US foundation models’ boundary-breaking capabilities, dispelling worries about “AI stagnation.”
But this doesn't mean Chinese players are left behind. Goldman Sachs has observed a clear and resilient “dynamic catch-up” cycle:
Every time US models make a major version leap (Step-up), Chinese AI models typically follow within 3-6 months, quickly closing the gap through technological iteration. This “lead-follow-lead again” rhythm has become the new normal in US-China tech competition.
Goldman also notes Chinese vendors’ resilience comes from their unique “China speed” and open-source ecosystem. Giants like Xiaomi, Tencent have open-sourced their high-performance models; reports say 80% of China’s AI startups are using these open-source models.
Additionally, Chinese models are highly aggressive in cost control: for example, Kuaishou’s “Kling” video generation model is priced significantly lower than global peers, building a moat through extreme value-for-money at the application layer.
Value Depression?
On valuation issues that concern investors, the Goldman Sachs Ronald Keung team analyzes:
“We are not yet in an AI bubble.”
The team believes the current China AI sector has not entered bubble territory. Tencent and Alibaba’s 2026 expected P/E ratios are 21 and 23, respectively—lower than Google (24), Amazon and Microsoft (28-30).
This article is from WeChat Official Account “Hard AI”. For more cutting-edge AI news, click here
Risk Warning and DisclaimerThe market carries risks; investment should be made cautiously. This article does not constitute personal investment advice and has not considered individual users’ specific investment objectives, financial situation, or needs. Users should consider whether any opinions, views, or conclusions in this article suit their particular circumstances. Investments based on this article are at your own risk. ```