Currently, there is no clear data or evidence proving that CAPEX has ended. I attended an offline discussion, and B2B companies specializing in organizational AI adoption have yet to convince enterprises and have not demonstrated profitability. Instead, more obvious facts are right in front of us:

AI adoption remains very low!

1️⃣🇺🇸 In the US, only about 6% of enterprises have AI job penetration, but within that 6%, the top 1% contribute 90% of AI-related positions, which are highly concentrated in large companies.

2️⃣🇨🇳 Microsoft's Q1 2026 data: Only 16% of China's working-age population actively uses AI for over 90 minutes per month.

3️⃣ Doubao, Kimi, and GLM have not yet reached Claude's level.

4️⃣ AI adoption in traditional large enterprises in China, the US, and even the world has not yet begun.

5️⃣ Improvement in US-China economic relations: hardware imports and domestic hardware substitution go hand in hand, increasing hardware sales.

6️⃣ AI arms race. This means CAPEX still needs to be further expanded.