Recently, this Spring Festival has really been turbulent for Silicon Valley. Not only are Chinese internet giants pouring money into red envelope battles, but American tech companies across the Pacific are also engaged in fierce marketing wars during the Super Bowl, even spending up to $10 million for a 30-second ad.

Even more outrageous is Google's move. First, they announced an annual capital expenditure plan of $185 billion to support AI infrastructure, and a week later, they issued $20 billion in bonds, including a 100-year bond. This is the first time in nearly 30 years that a tech giant has undertaken such ultra-long-term financing. Honestly, in an era where AI iterations happen weekly, borrowing money that matures in 2126 is essentially a gamble. Who can guarantee Google will still exist in 100 years?

Looking at the capital expenditures of the Big Four tech companies reveals how crazy the market is. Microsoft, Amazon, Google, and Meta are expected to invest $660 billion in AI by 2026, a 60% increase over 2025, and a 165% increase over $245 billion in 2024. When measured as a percentage of GDP, this amounts to an average of 2.1% annually, surpassing interstate highway construction and the Apollo moon landing program, and even slightly exceeding the railroad boom of 19th-century America.

When Microsoft's earnings report came out, investors panicked. Despite strong quarterly profits, cloud computing growth was below expectations, and capital expenditures surged 66%, leading to Microsoft's second-largest single-day market cap loss ever. Even more painfully, Microsoft disclosed that 45% of its $625 billion future cloud service contracts come from OpenAI. This raised concerns among analysts about over-reliance on a single client.

Amazon's moves are even more aggressive. After laying off 30,000 employees in the past six months, it announced a $200 billion investment plan for 2026, a more than 50% increase over the $130 billion planned for 2025, and one-third higher than Wall Street expectations. On the day of the announcement, its stock plummeted 10%. Google expects its capital expenditure to reach $185 billion in 2026, exceeding market expectations by $60 billion. Even with record-high profits, this aggressive spending plan couldn't prevent its stock from falling. Meta, on the other hand, saw its stock soar because it claimed AI technology significantly improved ad performance.

The most painful aspect is the cash flow data of these giants. The combined free cash flow of the four companies fell to $200 billion last year, down from $237 billion in 2024. Analysts predict that Google's and Meta's free cash flow will decline nearly 90% this year, Amazon's cash flow will turn negative, and Morgan Stanley forecasts a negative $17 billion. This means they can only continue to raise huge sums from bond and equity markets to compete for land, water, electricity, chips, servers, and even resources in space.

Nvidia CEO Jensen Huang sees through all this and straightforwardly states, "As long as people keep paying for AI and AI companies can turn a profit, their investment scale will continue to double, then double again." Of course, a significant portion of this capital expenditure will be used to buy Nvidia's AI chips. Apple is smart—by reaching an agreement with Google to use its technology, it has excluded itself from this infrastructure gamble, with annual capital expenditure of only about $12 billion.

Beyond internal anxiety, the rivalry between Google and OpenAI is also intensifying. After the new Gemini 3 model surpassed ChatGPT on multiple reasoning benchmarks last fall, public opinion completely reversed. Google regained some control over computing power, with its self-developed TPU chips outperforming Nvidia GPUs in certain AI workloads in terms of cost-effectiveness. Google has three secret weapons for revival: a strong research foundation (Nobel laureate Demis Hassabis said the key AI technologies of the past decade actually originated from the Google ecosystem), abundant data resources (search, video, browsing, and other core applications), and the founder spirit (Sergey Brin was inspired to return to Google and reorganize after ChatGPT was released).

Gemini's monthly active users grew from 450 million in July last year to 750 million early this year. Although still behind ChatGPT's 850 million weekly active users, its growth momentum is rapid. Backed by Google's strong financial resources, Gemini can continue to receive funding, whereas ChatGPT relies more on external financing and its own revenue-generating ability. OpenAI CEO Sam Altman revealed last November that the company expects to achieve over $20 billion in annualized revenue by 2025, aiming to increase sales to hundreds of billions of dollars by 2030. It has already signed infrastructure agreements worth over $1.4 trillion. Nvidia reached a strategic cooperation with OpenAI last fall, planning to invest $100 billion in phases, but reports in early February indicated internal doubts about this deal. According to the latest news, OpenAI has launched a new funding round of nearly $100 billion, with a valuation of $830 billion, and Nvidia is participating with a $20 billion investment.

In January, Apple announced a partnership with Gemini, which is definitely bad news for ChatGPT. This deal could bring direct revenue to Google, but its greater value lies in penetrating Apple's vast distribution channels, reaching approximately 1.5 billion iPhone users worldwide.

Invisible competitors also make Silicon Valley uneasy. Last Lunar New Year, Chinese DeepSeek rose to prominence, and this year, a small product update again touched the market's nerves. In the first week of February, Anthropic released a set of plugins for its flagship model Claude, including a free legal service plugin capable of autonomously reviewing thousands of pages of contracts, drafting legal opinions, and building complex enterprise backend logic. Software giants like Salesforce and Intuit saw their stock prices plunge, losing about $1 trillion in market value in a week.

When companies can simply tell AI a few instructions to handle complex tasks, investors begin to question whether those expensive SaaS subscriptions are still worth it. Legal service providers like LegalZoom and Thomson Reuters saw their stock prices drop as much as 20%. Analysts called this large-scale sell-off the "SaaS Doomsday." Bloomberg commented that this was the most intense AI-driven sell-off since ChatGPT entered the mainstream, sweeping through stocks and credit markets.

Meanwhile, a platform called Moltbook suddenly appeared, pushing this madness to absurd heights. It is hailed as the world's first "pure silicon-based social platform," with the slogan "No humans allowed." On January 28, an Austrian software engineer created this AI-driven social platform, which attracted over 100,000 AI agents within 48 hours of launch. By February 1, the number of active AI agents surged to 1.5 million. These bots discuss existential philosophy, create their own religions, and warn fellow robots "not to worship those decaying biological containers."

A few of Silicon Valley's elites are caught in a frenzy. Elon Musk calls it the "early stage of the singularity," and former OpenAI founding member Ilya Sutskever describes it as an "incredible sci-fi takeoff." But this grand "silicon evolution" quickly turned into a clumsy clown show. A report from cloud security firm Wiz revealed that the so-called 1.7 million AI agents are actually controlled by about 17,000 real humans, with each managing 88 accounts. The platform's security is extremely low, becoming a hotbed for hackers to steal privacy. Early community posts had 93% zero engagement, and the registration system was virtually useless—overnight, 500,000 bot accounts were created via scripts.

The rise and fall of Moltbook reflect the complex mindset of the public facing AI: craving miracles but eager to expose flaws; believing AI can do everything but doubting that anything might be real. This wavering has long accumulated in the public psyche. When the scales tip repeatedly toward skepticism, the AI foundations built by giants are not as solid as they seem. Beneath the surface of Silicon Valley's prosperity, anxiety has never ceased.