NVIDIA +9%... A major counterattack of the "Jevons Paradox"

The day after Deepseek trading; QQQ +1.5% + NVDA +9% + Software +2.8%, a rather impressive rebound. Goldman Sachs' trading desk directly named this counterattack "Jevons Paradox";

First, let me wish everyone a Happy New Year, good health, and safety!

Some sentiments and comments observed last night;

1/ Yesterday we mentioned that short sellers had many doubts about the "Jevons Paradox," and hedge funds found it difficult to catch this "falling knife" on the first day + many long-only investors were busy reducing positions before digesting the information; today, more sell-side analysts came out to support computing power + "Jevons Paradox"; for example, during the MS conference call yesterday morning in the Eastern US, there were over 4,000 participants... of course, foreigners don't celebrate the New Year; but even during the Spring Festival, there were over 2,000 friends in China urgently learning the details related to Deepseek. Foreigners only know we have many engineers... they have no idea how competitive other industries are...

2/ Before NVIDIA made its move in the morning, we already saw the AI software narrative line performing quite strongly; regardless of the fluctuations after semiconductors, at least the market has agreed on model technology iteration -> lower inference costs -> more data consumption -> lower AI agent prices -> more demand -> better margins + more revenue + faster technology adoption; CRM, NET, FSLY, SNOW all performed well;

3/ Two comments that were discussed by many yesterday, one from Sam; the other from Andrej Karpathy (former OAI/TLSA); regarding computing power.

4/ A few more words about software; Cloudflare, CrowdStrike, Confluent, Monday.com, MongoDB, Zscaler, Snowflake, etc., all saw increases of 7-12%; from Goldman Sachs' trading desk observations, the previous trading day's strength in software stocks reflected more technical factors (such as short covering), while yesterday's trading day saw a more substantive rise; some investors seem to hold a more open attitude towards new growth points in the TMT sector 5/ Alibaba Cloud's QWEN has also been released, and BABA seems to be reacting a bit; previously, Chinese internet giants have never enjoyed any premium related to AI; there may be opportunities for re-rating to watch.

6/ What to pay attention to next? FOMC + major earnings (pre-market ASML + SMCI, post-market MSFT, META, NOW, TSLA);

Finally, here’s a discussion I saw about Capex / Deepseek, which is well-written and resonates with me; this wave of AI investment is funded by tech giants with deep pockets (their FCF % is more than enough to support this expenditure; and they have been making money effortlessly for so many years, the discussions in previous years were about how these giants would spend this money, which is why buybacks were so prevalent); during the internet bubble, these giants had not yet emerged, and the investment cycle at that time mainly came from venture capital (VC), whose money needed to be accountable to the subsequent LP pressure, and the companies they invested in did not have strong balance sheets, making it easy for the investment cycle to break under a big wave.

The following original text comes from a TMT team, and I translated it directly using Deepseek;

If we compare "AI" to the internet (both Altman and Bezos likened it to transistors or electricity in their Dealbook interview, but "internet" may be the most common analogy)... then this technology will certainly spread rapidly to the general consumer and businesses, and the cost of infrastructure will no longer be the focus of investors; what truly matters is the innovation and business models brought about by this "new technology." The internet significantly boosted GDP and productivity, and if AI has a similar effect, we will surely witness an economic boom (there was even a budget surplus during Clinton's presidency...).

But let's temporarily set aside macro narratives like "open source," "internet," and "benefiting the public," and assume you are running a physical business that aspires to navigate through technological cycles and remain evergreen... Would the developments of the past week lead you to cut total AI investment?

The internet cycle has not been friendly to most companies. The bursting of the bubble destroyed reputations, hedge funds, careers, and capital, with the most notable feature being that the winners are few and far between. While human civilization/GDP/society as a whole benefits, we are trading companies, not "society." What common characteristics do the survivors who navigate bull and bear markets share? They double down on investments against the trend, continuously investing in hard technology, physical assets, production facilities, infrastructure, and especially human capital during tough times. No survivor has succeeded by being stingy.

The logic that those who are determined to succeed will now tighten spending is indeed far-fetched. The true survivors of the internet bubble are only Amazon. This company is known for "your profit is my opportunity," completely overturning the perception of e-commerce as "high margin" and "light asset." They acquired planes, trucks, and warehouses, employing 1.6 million people. Bezos was not as sought after by capital in the early days as he is now; his investment cycle was full of controversy and attracted a lot of external criticism. The key to Amazon's success lies in its free cash flow and self-sufficient funding, which allowed it to weather the capital winter Today, all tech giants clearly possess this strength.

Born in the financing winter after the internet bubble, Google is famous for its garage startup origins, but its development trajectory is contrary to "controlling expenses." Google spends lavishly on building data centers, hiring high-paid talent, creating expensive campuses, making numerous acquisitions (including talent acquisitions), and investing in fundamental software and hardware. Wall Street has bet twice that its CFO would tighten the purse strings, and criticisms of its spending have persisted for years, but history has proven how ridiculous such short-sighted concerns are. Google's technological achievements are severely underestimated (just ask a child under 12 if they know about Microsoft).

There is almost no evidence that tech giants (the real financial backers) will reduce investments in any area due to this incident. The most ambitious startups will not slow down either. OpenAI is a special case: as the most valuable leading brand in this movement, if operated properly, it could become the Google of 2004. Its high-risk, high-reward nature makes its large-scale investments unsurprising, and its model releases always attract far more attention than its peers (which is crucial for both the brand and the AI movement).

It is indeed surprising to adopt a capital preservation/contraction training budget strategy at the beginning of a multi-year technology cycle.

Risk Warning and Disclaimer

The market has risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article align with their specific circumstances. Investing based on this is at your own risk