From Tencent and Circle: Looking at the Simple and Difficult Questions of Investment

Author: Alex Xu

The narrative around AI continues to ferment, yet the recent performance of related stocks varies widely, with some experiencing a summer surge while others feel like they are in a winter chill.

A representative of the former is Circle, the issuer of the largest compliant stablecoin USDC, while a representative of the latter is Tencent.

There are many reasons for Circle's rise, such as its previous drop from 300 to 49, which provided a deep enough correction to accumulate rebound momentum. Additionally, USDC's scale has recently seen a rebound, reaching new highs again.

However, the biggest driving force may still be the explosive popularity of openclaw, which has shocked people with the wave of Agents. This tangible impact has ignited interest in stablecoins deemed very suitable for Agent settlements, and Circle is currently the only pure-blood stablecoin among publicly listed companies worldwide.

The reasons for Tencent's decline are thought-provoking. Previously criticized for lagging in the AI era, it has continuously corrected from a high of 680. Following a recent earnings report that showed strong performance but also indicated a "reduction in buybacks and increased AI investment," it has not only failed to rise but has instead fallen further. Market concerns about its AI investments have shifted from "insufficient" to "excessive."

So, in the AI era, which company has more certainty, Tencent or Circle?

Charlie Munger once said that investors should avoid difficult problems, and his exact words were:

"One of our secrets is not to try to understand too much... I constantly categorize things as 'too difficult.' Occasionally, an easy decision comes along, and I make it. That's my system: most things fall into 'too difficult,' and only a few obvious simple decisions I will handle immediately."

So, now, which is a simple problem on our investment table, Tencent or Circle, and which is a difficult problem?

Let's start with Tencent.

Currently, the Yuanbao AI in WeChat is actually playing an experimental role for future WeChat's own Agent:

It can already accomplish most of the high-frequency tasks that ordinary people use AI for, namely: queries and Q&A.

However, Yuanbao can currently only interact with content-type information within WeChat, such as text and image information, video account content, public account articles, etc. At this stage, it cannot directly interact with mini-programs.

But it is not hard to predict what WeChat's native agent might look like: through natural language dialogue, we can directly instruct WeChat's Agent to call mini-programs to perform various tasks, including ordering takeout, booking hotels, summarizing videos and recordings, reminding schedules, and reporting work.

Mini-programs themselves constitute a huge application ecosystem, with most of the commonly used apps having mini-program versions: Didi, Meituan, Pinduoduo, McDonald's, HuoLala, WPS, and even Alibaba's Xianyu and Ele.me.

More importantly, WeChat's agent has strong system-level permissions regarding mini-programs, similar to (or even stronger than) Apple's software for the iOS ecosystem, and mini-program development must adhere to WeChat's specifications. WeChat's agent will not be directly banned by major apps like Doubao's mobile assistant, which is kept outside the walled garden, but will interact with various mini-programs at a system level. Various mini-program apps will inevitably embrace WeChat's official agent to provide better services to users.

Compared to openclaw, which remains popular but has left many early adopters frustrated and caused a large number of ordinary users to give up, WeChat's Agent, aimed at 1.4 billion users, will undoubtedly have better usability and practicality. Only then will it truly represent "mass-level Agent promotion." This form may not be as geeky or cool, but it is a product usable by 90% of the public.

The launch of such an Agent product at some point in the future is highly certain. However, before it happens and before everyone sees it, with Tencent's price still falling, Mr. Market is stingy in giving it a valuation.

But for value investors, Mr. Market is what we need to take advantage of.

We should invest in those high-probability, even inevitable events that have not yet occurred.

In the face of the highly uncertain AI era, is Tencent's operational certainty really that low? I think it's not too bad.

At this moment, is buying/holding Tencent a simple problem or a difficult problem?

I believe it is far simpler than what the market sentiment reflects.

Now, let's talk about Circle:

When looking at some things over a longer term, the certainty is actually quite vague.

Will Circle's future business scale smoothly grow to several hundred billion or even several trillion?

Optimistic bullish voices believe it can, and their logic is:

1. AI is the future, and future agents will be the main users of the internet, dominating a massive scale of online transactions.

2. AI transactions will require stablecoins settled on the blockchain.

3. USDC is currently the largest compliant stablecoin, and the demand for AI settlements will directly drive the growth of USDC's scale.

4. Stablecoins have strong network effects, and the expansion of scale will self-reinforce.

5. USDC's scale will grow in sync with the scale of AI transactions to several trillion or more.

This chain of reasoning seems tightly knit, but several important links actually have considerable uncertainty:

1. Will future agents be the main users of the internet, dominating a massive scale of online transactions?

Looking at it over the long term, this high certainty is overestimated.

2. Will AI transactions require stablecoins?

Not necessarily. The first point is visible to everyone, and existing payment infrastructure service providers and internet and AI giants that control users are also pushing their own standards, many of which are not blockchain-based, such as Google's UCP protocol aimed at agentic commerce.

3. Is USDC the largest compliant stablecoin?

It is currently, but Tether also has significant power and is attempting to become compliant, with considerable political resources. There are also more eager financial groups and payment providers like PayPal.

4. Do stablecoins have strong network effects?

In the face of ruthless agents, the differentiation of stablecoins will be compressed. As long as they can be used for payments, agents will choose the ones with the least friction and cost, regardless of whether it is USD1, USDT, USDC, or a stablecoin issued by a bank. You say USDC currently has more merchant-side resources? Visa, Stripe, and other payment channel providers have a merchant network and channels far greater than Circle's, which can also help other stablecoins integrate. As long as the settlement is convenient and friction is low, merchants do not care which stablecoin they receive. The network effect of stablecoins may not be that strong in the agent-led payment era.

5. Will USDC's scale grow in sync with the scale of AI transactions to several trillion?

When the assumptions of 2-4 all have high uncertainty, combining them to derive conclusion 5 makes the uncertainty even more vague.

In the face of the highly uncertain AI era, is Circle's operational certainty really that high? I think it's not too bad.

At this moment, is buying/holding Circle a simple problem or a difficult problem?

I believe it is far more difficult than what the market sentiment reflects.

This is just one perspective, for reference only.