EP687 | 🐧

New designated-stock rules bring short-term trading into a phase of mutual probing

After Taiwan's designated stocks switched to matching once every two minutes, the market is still feeling out the new rules. The host observed that fake orders intended to mislead other traders during the session appear to have become more frequent. Such orders may not be intended to execute at all, but rather to create an illusion and influence an opponent's judgment. For ordinary retail investors, the most common result may simply be a price difference of a few ticks; the people who truly need to calculate the details of the rules are those trying to engage in short-term speculation.

The host compared the situation to two giants throwing punches at each other while bystanders are caught in the wind of the blows: large players may not be targeting retail investors, who are simply standing near the two sides of the contest. The deeper problem is that the market once operated on an implicit understanding—that good revenue performance should normally lead to a rising share price. Now, even when revenue far exceeds expectations, the stock may be sold first. Participants then begin anticipating what everyone else will do, eventually creating a situation in which everyone rushes to act first.

Once trust is lost, good news no longer guarantees a rise

When the market repeatedly sees good news followed by selling, investors begin to wonder whether the news had already been priced in. Even friends who originally liked the fundamentals may sell first the next day to avoid being caught by another reversal. The host believes this makes the market look more like a range-bound market, increasing the difficulty of short-term trading. The market may need to make new highs again before it can rebuild the consensus that good news should push share prices higher.

He remains relatively optimistic because some groups are gradually taking shape, with many small- and mid-cap themes and growth stocks beginning to reach record highs. However, if investors track every fluctuation with an overly short-term mindset, they may sell out of their holdings before the real move begins. In the current market, a more reasonable approach is to accept that participants will probe one another in the short term, while taking a longer view of price patterns and whether capital flows continue to improve.

Google's adjustment may be moving model risk outside the company

Over the weekend, the market discussed Google's personnel changes and the fact that the Gemini update fell short of expectations, prompting speculation about whether Google was withdrawing from the frontier-model race. The host does not believe that a single personnel arrangement is enough to prove Google has abandoned state-of-the-art models. Google is a conglomerate made up of multiple businesses, so short-term resource adjustments may represent a change in how it exercises control rather than a halt in investment.

The change at Google$GOOGL looks more like bringing DeepMind's previously relatively independent dual-track structure back under parent-company management, with its successor reporting directly to CEO Sundar Pichai. The market may also interpret growth in the cloud business as meaning that computing capacity is being rented out because Google's own software is not strong enough. But the host notes that this is not necessarily a bad thing: if Google moves the highly uncertain investment in frontier models outside the company while retaining cloud, computing capacity, and service revenue, it may be able to spread out both the risk and the capital expenditure.

This transformation also creates a valuation issue. If the market no longer views Google as a major option on winning the frontier-model race, the higher price-to-earnings multiple previously granted on the strength of AI expectations could come down. But if capital expenditure converges more quickly and cash flow turns positive sooner as a result, the valuation could instead rise. The host is therefore in no hurry to declare a winner or loser, and will watch whether the share price continues to make new lows after the personnel news. If it does not break down and resumes rising, the market may be accepting a new infrastructure narrative.

The biggest variables for frontier models are the value chain and Chinese challengers

The frontier-model race is currently being fought fiercely by companies such as OpenAI and Anthropic, while other major cloud providers are also investing. Even if the Western market ultimately produces one or two leaders, that does not mean they can monopolize value worldwide. Chinese companies may catch up through lower prices, subsidies, or rapid replication. If competitors cannot be completely shut out, some of the extensive R&D成果 produced by the leaders may be captured by low-cost services.