Alphabet Earnings to Test AI Bubble Fears, Soothe Chip Anxiety

Alphabet Inc.'s earnings report on July 22 is set to become a key market test as investors debate whether the rally in artificial intelligence and semiconductors has reached its peak.

Debate over the end of the semiconductor upcycle has intensified in recent weeks. The central question is whether heavy AI spending by hyperscalers including Google, Microsoft, Amazon and Meta Platforms can translate into real revenue and profit growth. If massive capital expenditure fails to deliver sufficient sales and earnings, Big Tech could scale back investment, stoking concern that demand for AI chips such as graphics processing units and high-bandwidth memory will slow.

The arrival of Chinese AI model Kimi K3 has deepened those concerns. The model offers performance close to the top versions of Claude and ChatGPT at a relatively low price. Analysts say the implications are significant because a Chinese company unveiled a model with sharply improved computing efficiency even as US export controls on advanced semiconductors to China remain in place.

That could pose a threat not only to US AI companies but also to the chip industry. The core bullish case for semiconductor stocks has been that developing and serving more powerful AI models will require ever more GPUs, HBM and data centers. But if efficiency improvements allow similar performance with fewer computing resources, hyperscalers may need fewer high-end chips than previously projected.

Alphabet has guided for capital expenditure of $180 billion to $190 billion this year, about double last year's $91.4 billion. If it maintains that investment plan in this earnings release and confirms strong growth at Google Cloud and demand for AI services, anxiety over semiconductor stocks may ease. If cloud growth slows or revenue and profit gains remain modest relative to AI investment, the market could interpret the AI industry's spending to date as overinvestment.

Recent market unease has also been fueled by delays to the launch of Google's next-generation AI model, Gemini 3.5 Pro. Some key capabilities, including coding performance, have reportedly fallen short of internal targets. By contrast, China has produced Kimi K3, a model from a company with restricted access to advanced chips that still improved computing efficiency.

Lee Hye-in, Hankyung.com reporter hey@hankyung.com