Hang Seng Index up 0.86%, Hang Seng Tech Index up 1.65%
MiniMax's stock price plunged over 15% in a single night, marking its worst performance since going public—and just as it's racing to list on the STAR Market. The scene is strikingly reminiscent of a tech upstart shouting "the future is bright" while secretly wiping away sweat. The AI industry bubble might just be quietly spilling over from these numbers.
Analysis
MiniMax's stock price plummeted over 15% in a single night, setting its worst daily record since listing—right as it's pushing for a STAR Market debut. This scene looks remarkably like a tech darling yelling "the future looks promising" while nervously breaking into a sweat. The AI bubble may indeed be quietly leaking through these figures.
Look at this trending list: both Zhipu and MiniMax are rushing to return to the A-share market. Why? Overseas valuations might be cooling, while the domestic market still seems to offer policy tailwinds and retail investor enthusiasm to rely on. But is returning to the A-share market truly a cure? The STAR Market story revolves around "hard tech," yet these large language model companies burn through money like water, their profit models as hazy as flowers seen through fog. Are investors buying into the technology, or the grand narrative of a "Chinese OpenAI"? When technological barriers are quickly leveled, how long can the story hold up?
What’s more ironic is that Silicon Valley giants just staged a "delicious reversal" (真香). After burning billions on GPUs and stockpiling computing power, Google and Microsoft are now restricting internal employee token usage. It’s almost like a large-scale performance art: first dramatically declaring "unlimited AI computing power," then sheepishly reminding employees to "use it sparingly." Cost pressures have finally punctured the dream—turns out AI isn’t magic, it’s a business. Every token generated burns real money. When commercialization falls short of expectations, R&D budgets are often the first to be cut. Could this slow down the next wave of technological evolution?
Speaking of evolution, Microsoft’s "Skills self-evolution project"—which gained 3.3k stars in a week—sounds impressive. Training skills like training a neural network? The concept is sexy, but practical implementation still faces countless engineering pitfalls. Developer community hype can sometimes resemble a collective frenzy, and the number of projects that truly become productivity tools is likely less than one in ten. Are we falling into a new round of tech worship, ignoring the dull, slow, and meticulous foundational optimizations?
Then there’s the "26 billion USD all-Chinese AI programming company" chased by capital. Its valuation is shockingly high, as if programming is on the verge of complete AI disruption. Yet the reality is, programmers are still working 996 schedules fixing bugs, and AI-written code still requires human review. Behind the high valuation lies either investment firms' greedy bet on "AI replacing everything" or a misjudgment of real market demand? When the tide recedes, survival for these companies will likely depend not on PPTs, but on actual customer payments and profits.
Looking back at Hyundai Motor’s sales decline, it may seem distant from AI, but the core is similar. Just as traditional industries fret over supply chains, isn’t the AI industry equally anxious about its own supply chain—computing power, data, and talent? MiniMax’s stock plunge and listing dream are a microcosm of this anxiety: the urgent need for capital infusion to stay alive, while facing the cold scrutiny of capital markets.
The AI narrative is undergoing a subtle disenchantment. From "changing the world" to "computing costs," from infinite imagination to returning to financial statements. This isn’t a bad thing—at least it makes us see that even the coolest technology must stay grounded. Still, as every company rushes to cash out or raise funds, how much patience and reverence for the technology itself remains?
Disclaimer: The above content is generated by AI and is for reference only.