每日：尽管科技股波动，AI基本面依然稳固

Video: Top of Mind APAC | Min Lan Tan on Asia IT's 342% profit surge (6 mins)Video: Market Playbook | Why commodities, including copper, are attractive (5 mins)Video: CIO Monthly | Kiran Ganesh on market drivers, scenarios, and opportunities (4 mins)Podcast: How AI may complicate inflation signals, on Apple and Spotify (6 mins)

Tech stocks have come under pressure over the past week as higher long-dated Treasury yields weighed on sentiment. Ongoing concerns over the sustainability of AI capex and the circularity of certain AI financing deals have also kept investors on edge. The Nasdaq has fallen 1.9% over the past 10 days, while the Philadelphia Semiconductor Index declined 8.2%.

NVIDIA’s earnings, due later today, could create further volatility, with investors looking for a robust earnings beat and forward guidance that signals sustained demand for AI chips.

Without taking a view on any individual company, we retain our conviction in the broader AI growth story and believe it remains a key driver of our positive market outlook.

AI demand continues to see robust growth. Consumption of tokens, the basic units of AI computing, remains one of the most reliable ways to gauge AI demand, and the latest OpenRouter data indicate 24% week-over-week growth in token volume. Over the past year, token consumption has risen more than 30-fold. Additionally, while token pricing has declined this year, recent signs of stabilization suggest that demand growth is more than offsetting ongoing efficiency gains. As the use of AI agents expands across industries, we believe robust demand should continue to support rising AI capital expenditure. We project total AI capex to hit USD 1.2tr in 2027, up 33% from our estimate of USD 900bn this year.

Monetization has also accelerated. Higher capex alone is no longer sufficient to sustain the AI trade, but evidence on monetization is encouraging. Major hyperscalers reported average cloud revenue growth of 48% year over year for the second quarter, well above consensus expectations and representing a further acceleration from previous quarters. We anticipate this acceleration in cloud revenue will con­tinue, forecasting growth of 58% in the current quarter. While the risk of investment running ahead of returns remains, Amazon CEO Andy Jassy recently suggested that the AI investment cycle reflects a timing mismatch rather than a structural drag on economics. He added that data center capital can support revenue generation for more than 30 years. Such confidence in the potential for attractive long-term free cash flow and returns on invested capital should help support the AI investment cycle.

Valuations have become more appealing, and earnings growth remains strong. In our view, the recent tech volatility has created increasingly attractive entry points for select companies. For example, a broad basket of key semiconductor companies now trades at around 22x forward earnings, below its average of 24x since the launch of ChatGPT in 2022, and well below the recent peak of 33x. More importantly, earnings growth remains robust, and earnings expectations continue to increase. Market consensus now projects earnings growth of 92% for key semiconductor companies this year, followed by 40% growth in 2027. For Nasdaq 100 companies, the consensus forecast is for earnings growth of 38% this year and 26% next year.

So, against this backdrop of solid near-term fundamentals but persistent risks, we recommend that investors hold a selective and diversified allocation to tech stocks. We continue to favor the enabling layer of the AI value chain, including semiconductors and cloud computing, but investors should also consider building exposure to more defensive tech segments such as payment networks, data center REITs, and select smartphone makers.