$ZKP and $ETH's recent performance in the AI sector has indeed attracted quite a bit of attention. Let me break down three interesting new projects:

First, $ACT. This project combines oracles and AI, which is quite a novel concept. Using on-chain data as input sources for AI models is a good idea. However, in reality, these types of projects heavily rely on market sentiment in the early stages, leading to sharp fluctuations. If there are no actual commercial partnerships or application implementations later on, the hype tends to fade faster than expected.

Next, $IO. This one is more practical. With so many idle GPUs worldwide, consolidating them onto a platform for on-demand rental makes sense. The demand is real, and the model is straightforward. Price trends are likely to be more stable because they are supported by genuine supply and demand. If you see growth in node numbers and large clients signing agreements, that’s usually a bullish signal.

$NIL mainly focuses on combining privacy computing with AI, where multiple parties collaboratively train models under encryption protection. The technical barriers are high, so a rapid surge in price is unlikely in the short term. However, once news of collaborations with leading enterprises or institutional-level partners emerges, it often attracts long-term capital, and the stock price may face re-pricing.

To summarize: if you prefer short-term trading, you can focus on the sentiment around $ACT; if you want relatively stable returns, the fundamentals of $IO are more worth researching; $NIL is a technical long-term bet. The core advice remains the same—true profit logic always lies in in-depth research, not following the crowd to chase gains or sell in panic.