From AI Hype to AI Proof

Semiconductor stocks extended their recent decline on Friday (17 July 2026), marking the sector’s weakest week in more than a year. The Philadelphia Semiconductor Index (SOX) recorded its steepest weekly decline since March 2025 and closed more than 20% below its June record high, officially entering bear market territory. Nvidia, Broadcom and Qualcomm all ended the week lower as investors continued to pull back from one of the market’s most crowded trades.

At first glance, it looked like another AI sell-off. But, when you take a bird’s eye view, the bigger story is that investors are becoming more selective about the companies driving the AI boom. In many ways, this wasn’t a new story at all. It was the market continuing a conversation that began months ago.

Back in May, I explored how the world’s largest technology companies were committing unprecedented sums to AI infrastructure and why those investments would eventually need to generate meaningful returns. A few weeks later, I argued that markets had stopped rewarding AI spending announcements and had started demanding evidence that those investments would justify increasingly expensive valuations.

Friday’s sell-off appears to reinforce that view. For almost two years, investors rewarded almost every company linked to AI infrastructure. That pushed valuations to levels where expectations became exceptionally high. When that happens, good earnings alone are no longer enough. Companies need to outperform forecasts and convince investors that today’s spending will translate into tomorrow’s profits.

That was evident this week when TSMC reported a 77% increase in quarterly profit, beat market expectations and raised its capital expenditure outlook. Under normal circumstances, that would have supported semiconductor shares. Instead, the stock still fell as investors focused less on the results themselves and more on whether future growth can continue to justify elevated valuations.

Another factor weighing on sentiment was the launch of Moonshot AI’s Kimi K3 model. The announcement renewed questions about whether increasingly capable and lower-cost AI models could eventually reduce the pricing power of some US AI leaders. It wasn’t the primary reason behind the correction, but it added to an environment where investors were already reassessing stretched valuations.

Importantly, there is still little evidence that the AI investment cycle is slowing. The major cloud providers continue to invest heavily in AI infrastructure, and demand for advanced chips remains robust. What appears to be changing is not the long-term outlook for AI, but the price investors are willing to pay for future growth.

That distinction is crucial to note. Markets don’t move purely on fundamentals; they move on expectations. The first phase of the AI rally was driven by excitement over what the technology could become. The next phase seems to be driven by execution.

For investors, that raises a much more important question than whether AI remains the future. The focus is shifting to which companies can actually convert billions of dollars of AI investment into sustainable earnings. The AI story hasn’t disappeared, it has simply matured. The market is no longer rewarding companies for talking about AI. It wants proof that AI can generate lasting profits. That is likely to define the next chapter of the AI trade.

More from AI series:

1. AI is moving faster than the global economy can adjust
2. The AI trade is entering its next phase: prove it
3. $650 billion and counting: what big tech’s AI spending says about the market
4. AI in Finance: What It Can Do, What It Cannot Do, and Why Humans Still Matter

Sources

Reuters; Yahoo Finance; Nasdaq Market Data.

Disclaimer: This article is for educational purposes only and should not be considered financial advice. Always conduct your own research before making investment decisions.