Two Lessons from My Visit to Tencent Cloud (1 of 2) (Tech Strategy)

Posted on by

I recently visited the Tencent HQ in Beijing. The goal was to dig into what Tencent is doing in cloud, especially AI cloud.

Because in the past six months, Tencent has been making some big moves in AI and cloud.

  • Lots of big investments in LLMs.
  • A deal with DeepSeek.
  • A big push to take cloud international.

And, most recently, Tencent issued a press release about their work with Goto in Indonesia. They successfully transitioned GoTo’s entire cloud to Tencent. That’s as high profile a project as there is in Southeast Asia.

So, I went up to Northwest Beijing (near my old office at Peking University) to visit their big (and pretty new) facility. Below is a model of it (from the lobby).

It’s pretty great. When it opened in 2018, this was the largest office in Asia. Today it has about +6,000 Tencent employees (out of 100,000).

I have two main take-aways from the visit:

  1. Tencent Cloud is doing a “fast follower with advantages” strategy for AI. That is my description (not Tencent’s). They have been rapidly repositioning in AI and making big investments to catch up to the leaders. And they are leaning into their strengths (especially WeChat) as they do so.
  2. Tencent Cloud is really going international. Including WeChat. This is something we haven’t seen since 2012-2015.

I’ll go into point 1 in this article. And do the other in another article.

A disclaimer. All of this is my summary and interpretation. None of this is Tencent’s explanation or words.

Take-Away 1: Tencent Cloud is Doing “Fast-Follower with Advantages” in AI

A few months ago, Tencent CEO Pony Ma was seated next to DeepSeek CEO Liang Wenfeng at a China entrepreneurs symposium in Guangdong. That was pretty symbolic. And it made for a good photo.

Tencent has done an important deal (partnership?) with DeepSeek. They have made DeepSeek’s open-source model available within their Yuanbao app, which is currently one of the top AI apps in China. Yuanbao is basically the ChatGPT of China (along with a couple of the contenders). But now when using it, you have two choices for which foundation model it runs on. You can choose Tencent’s Hunyuan foundation model model or DeepSeek.

Tencent adding DeepSeek to Yuanbao (and other products) was a pretty powerful move. With big benefits both companies.

First, Tencent and its main model Hunyuan were somewhat behind in China’s domestic AI models. At least according to performance and user experience rankings. There are lots of potential reasons for this. A common explanation is that Tencent has aligned its AI initiatives closely with its core businesses (advertising, gaming and social networking). As a result, they didn’t focus as much on building lots of foundational large models. In contrast, Baidu and Alibaba Cloud have been building models across the whole spectrum. Very aggressively.

If you buy that explanation, then the deal / partnership with DeepSeek makes sense. DeepSeek shocked everyone with its open source and cheap models. They are getting good adoption globally.

Tencent leapfrogged forward and can now offer knowledge models (good at search and chat) and reasoning models (DeepSeek).

In return, DeepSeek got a huge bump in usage. It got access to the vast user base and data of the Tencent ecosystem (especially WeChat). That means accelerated improvements in its models. DeepSeek also went into Tencent Cloud and Tencent Docs. And of course, into WeChat search, which is huge.

There is definitely some truth in this narrative. But I see it a bit different.

I think Tencent is doing their standard fast follower strategy.

Tencent has a unique position in the China ecosystem. It dominates messenger and social media. It has strong positions in most types of content. And it has other great products like payment, enterprise apps and gaming. It basically has unmatched visibility into what Chinese users are doing on their smartphones all day long. And it has tremendous distribution when it wants to roll out new products.

You can see Tencent taking advantage of its unique vantage point its M&A and investment strategy. When it sees users adopting new products, it makes small or controlling investments in those areas. And then it steers traffic to its partners and acquisitions.

Its unique vantage point also means it bide its time. New products and technologies means lots of innovation. With lots of failures. It takes time for the successful new products to become clear. When the winning product is clear (pricing, features, etc.), Tencent can aggressively copy it. Or buy a contender. And then push it in its ecosystem. Fast follower is a fantastic strategy. If you have the visibility and distribution to pull it off. Being first is usually too difficult.

WeChat was an example of Tencent’s fast follower strategy. Tencent was actually not the first mover in messenger. In fact, it was pretty behind when this took off. But it was able to come from behind quickly (by cross selling to its QQ and other products) and become the leader in messenger on the consumer side. DingTalk did better on the enterprise side (for now).

Anyways…

That’s what the recent AI moves by Tencent look like to me. They were a bit behind in 2024. But they are now moving up aggressively against the market leaders.

  • Tencent is definitely building lots of models. Hunyuan is proprietary, multimodal and a mixture of experts. But they are building across the board. They are doing a mixture of external and self-developed models. And they appear to be shifting from closed to open models. Like everyone else, they appear to be shifting towards open since they emergence of DeepSeek.
  • They are using a mix of internal and external products. Having Hunyuan (internal) and DeepSeek (external) is not surprising. Tencent likes to create internal competition when it launches new products. We saw the same approach when it launched search internally but also had a stake in #2 search engine Sogou (external). Tencent eventually acquired majority of Sogou.
  • They are also building data centers. Like everyone in this space, they are spending big. They spent 78b RMB in capex in 2024. Here’s a server from their exhibition space.

I Expect Tencent Cloud to Be One of the World’s Top 5 AI Cloud Companies

The other half of “fast follower with advantages” strategy is “with advantages”. Tencent are definitely leaning into their advantages, such as their huge user base, usage, data and product portfolio.

Unsurprisingly, Tencent is putting AI into all their products. Both the consumer and business facing, with a pretty even mix. That’s interesting. And they are embedding into their current apps (WeChat) as well as creating AI-native apps (Yuanbao).

WeChat is obviously the big product here. Yes, that means tons of users using AI search and generating lots of data and feedback (which makes the model smarter). It also means lots of merchant tools (think official accounts). And it means lots of advertising use cases.

Tencent’s 4 main GenAI use cases (thus far) appear to be:

  • Their ChatGPT-like AI assistant.
  • Enterprise tool.
  • Enterprise tool.
  • An AI coder for developers

That’s an interesting mix. It’s a big focus on consumers and other direct users with Yuanbao. That’s a lot like OpenAI. And they can definitely lean into WeChat with this. Note: In 2025 there are 1.38B people on WeChat.

But they also have a significant AI focus on Meetings and Docs. That’s WeCom, their enterprise service. That makes them look a lot like Microsoft. You can see them moving into industry specific AI models and tools.

Plus, something really interesting is Tencent’s focus on advertising use cases. They are the advertising giant of China and GenAI has tons of interesting use cases in this area. That makes them look like Meta. And advertising use cases can immediately drive revenue. How do you use GenAI to increase click rate? To increase revenue? Note: In 2024, Tencent revenue was up 8%. And Op profit was up 24%. Advertising meets AI is the area I am most interested in for Tencent.

That’s my current take. Overall, they are moving fast with a good strategy. Definitely a top 3 player to watch in China for AI Cloud. But they are also going international. I put them in the top globally now (AWS, Google Cloud, Azure, Alibaba Cloud, Tencent Cloud). Especially in Asia.

Final Point: The AI Cloud Digital Humans Are Pretty Cool.

These digital avatars / digital humans are just fun. And they have clear advertising functions.

They are already offering brands and merchants the ability to create digital humans as brand officers. Plus, you can now spin up lots of virtual influencers.  There are lots of other interesting applications like business chatbots, virtual bank officers and avatars giving tours at museums.

***

It’s hard not to be excited about Tencent and the other major China AI players.

The US is definitely better at ground-breaking research. But China is better at adoption and use cases. And at making things cheaper (which is very important). We have long seen this pattern in digital. And we can now see it again in GenAI.

Notice the current US focus on high-end computing infrastructure, using lots of the most powerful chips. While China is focusing on combining less powerful chips and targeting the low-cost use cases, especially on the consumer side. DeepSeek was also an example of this pattern.

AI agents are likely the big story of 2025. In both the US and China. I expect to see the same pattern.

But after that, I think the US and China paths might diverge. Chinese AI leaders will focus more on areas such as appliances (Midea, Gree), drones (DJI) and robotics (Unitree). Where China’s manufacturing scale creates an advantage.

That’s it for Part 1.

In Part 2, I’ll go into my other take-away: Tencent Cloud is really going international.

Cheers, Jeff

———

Related articles:

From the Concept Library, concepts for this article are:

  • n/a

From the Company Library, companies for this article are:

  • Tencent Cloud

———

I write, speak and consult about how to win (and not lose) in digital strategy and transformation.

I am the founder of TechMoat Consulting, a boutique consulting firm that helps retailers, brands, and technology companies exploit digital change to grow faster, innovate better and build digital moats. Get in touch here.

My book series Moats and Marathons is one-of-a-kind framework for building and measuring competitive advantages in digital businesses.

This content (articles, podcasts, website info) is not investment, legal or tax advice. The information and opinions from me and any guests may be incorrect. The numbers and information may be wrong. The views expressed may no longer be relevant or accurate. This is not investment advice. Investing is risky. Do your own research.

Leave a Reply