Written on the eve of Meta's earnings report, a brief discussion on market noise and independent thinking

This article was written on January 23rd and published on January 27th. I was too lazy to post it earlier... Just in time for the tech stock Black Monday triggered by Deepseek, I decided to write a bit more and share it. There are two days left until the earnings release of $Meta Platforms(META.US). This article is solely based on the author's personal investment reflections and experiences and does not constitute any investment advice.

$Meta Platforms(META.US) Meta is my largest holding, established six months ago at a price of 460, and I have never sold.

First, let’s talk about why I bought it.

In June, I bought a pair of Ray-Ban Meta smart glasses. After using them for a month, I really liked this little gadget—it achieved what Google Lab failed to do back in the day. This is the right product, and Zuckerberg’s vision was ahead of its time. Now, we can see major manufacturers imitating it. At the time, I considered that while every company is working on large models, chatbots are hard to monetize. After all, they need to be applied to real-world scenarios, or only hardware vendors will benefit. These glasses showed me Meta’s strategic foresight and execution capabilities, which are ahead of its peers. Additionally, AI significantly enhances the precision of Meta’s core advertising business. With over 3 billion daily active users across its app family, Meta already has a strong moat. Investing in AI further increases user stickiness and solidifies that moat.

Having explained why I bought Meta, I’d like to focus on the market noise, rumors, and how to navigate them.

Over the past six months, there have been many trading rumors about Meta. I’ll highlight three:

1. The presidential election

2. How to handle TikTok

3. The recent buzz around the domestic open-source model Deepseek

First, the presidential election.

Trump’s victory is a short-term negative for left-leaning Silicon Valley companies.

After the shooting incident, Zuckerberg clearly leaned toward the Republican side, quickly appearing on interviews to cozy up to Trump and even having phone calls with him.

The biggest concern was Trump’s statement that he would send Zuckerberg to jail if he took office. This is clearly noise. Trump is a businessman who wrote *The Art of the Deal*. His modus operandi is to talk tough but act softly—everything is negotiable.

The political impact of the election was effectively over once Trump announced he had a pleasant phone call with Zuckerberg. Some people say it sucks, but being a CEO requires these qualities. If you don’t like it, don’t invest.

Next, the TikTok issue.

Any hint of a TikTok ban immediately affects Meta’s stock price that day.

Especially with Trump’s flip-flopping—before his inauguration, he loudly advocated to "save TikTok." But I can clearly say this is noise. The fundamental impact of TikTok on Meta has already been priced in. The rest is just institutions leveraging the news for momentum.

Moreover, what does Trump gain from "saving" TikTok? With Zuckerberg already bowing to him, does the U.S. government really need to allow another hard-to-manage "Chinese spy app"? It’s purely for MAGA fans—Democrats want to ban your free speech and violate the Constitution, while I’m defending it. It’s all about votes. Remember, politicians don’t care about these issues; they only want votes.

Now, the main point of today’s article. I drafted this on January 23rd but procrastinated for days. Today, with Deepseek blowing up in the media, I want to reflect properly.

Meta just hit a new 52-week high during trading. Congratulations to those who bought the dip pre-market or held without selling—you made the right call.

I first learned about Deepseek from the image above, likely last Thursday night. I also checked the original post, which roughly said China’s open-source model achieved ChatGPT-level capabilities at an extremely low cost, causing panic among companies struggling to justify their massive spending.

At the time, I didn’t think much of it. But on Friday pre-market, Zuckerberg mentioned in an interview that Meta’s AI budget for 2025 would far exceed expectations at $6.5 billion, causing pre-market volatility and a brief semiconductor rally—though Meta still hit new highs during the session. However, semiconductors plunged on Friday, which I believe was already influenced by Deepseek.

Then, over the weekend, the media frenzy exploded. Deepseek topped app download charts in both China and the U.S., triggering a black swan event. Meta dropped as much as 6% pre-market but still reached new highs during the session.

I wasn’t worried at all, though some might call this hindsight. Here’s why:

First, this is market makers colluding with the media to shake out weak hands. Checking Twitter timelines, people were already praising Deepseek a week ago, and universities and big tech were scrambling to study it. Multiple academic institutions have replicated Deepseek’s success, proving this isn’t new news—it’s already been discussed and tested in the industry.

So why did the media only explode over the weekend? Because last week’s trading was dominated by Trump’s $500 billion AI strategy, which was more bullish for AI stocks.

The day before yesterday, Meta’s Chief AI Scientist Yann LeCun tweeted that Deepseek’s success is a victory for open-source over closed-source. I couldn’t agree more.

Deepseek’s success has made investors question whether so much computing power is necessary, triggering a short-term semiconductor crash and dragging down tech stocks.

For Meta, Deepseek is not a negative—only a positive.

Meta’s Llama is also an open-source model. This is progress and a win for the open-source community. Deepseek is a gift to the world, empowering small companies to venture into AI and democratizing AI for small businesses.

AI advancements directly enhance Meta’s core social media business and its ad revenue.

On a side note, this post and Zuckerberg’s interview show the company’s quick reflexes and competitive mindset.

Some might argue: "Deepseek did better than Llama with $5 million versus your billions—how is that not a negative?"

This is laughable. Big tech has money to burn—if not AI, what else would they research? Being on the right open-source path, even if wasteful, isn’t a problem. Deepseek accelerates AI progress and might even reduce costs—a huge positive.

Deepseek is also open-source. CEO Liang Wenfeng himself said their biggest constraint is computing power. Meta has more people, more GPUs, and can replicate Deepseek’s publicly available research. If they can’t catch up in two weeks, they might as well fire everyone.

Unfortunately, I didn’t practice what I preach—I sold BRK pre-market to buy NVIDIA instead of Meta... Regretting it now.

Now, let’s talk about semiconductors and computing power.

Here’s the Jevons Paradox (explained by ChatGPT):

The Jevons Paradox, proposed by 19th-century economist William Stanley Jevons, describes a counterintuitive phenomenon where improved efficiency in resource usage leads to increased overall consumption of that resource.

Specifically, as the efficiency of a resource (e.g., coal, energy) improves, per-unit consumption decreases, but the lower cost drives higher demand, potentially increasing total consumption. For example, more efficient coal use lowers prices, boosting industrial and household demand, ultimately raising total coal consumption.

Long-term, Deepseek is bullish for semiconductors. Short-term, the market can interpret it however it wants.

Liang Wenfeng himself said their biggest bottleneck is high-end computing power.

Straight to the conclusion:

If our goal is AGI and humanoid robots, today is a great buying opportunity.

Deepseek has achieved the current optimal solution for chatbots. If progress slows and stalls at chatbots/image generation, investor patience will wear thin, and the bubble will burst here.

Let’s encourage each other.