The recent volatility in global technology stocks has once again intensified this week. This is primarily driven by crowded positioning, with extreme leverage in markets like South Korea amplifying the swings. However, it's important to note that the recent decline in the AI sector is not broad-based; instead, it has narrowed to focus on the memory industry chain—including memory chips and semiconductor materials/equipment—following Micron's earnings report, which provided early validation of the sector's strength. Therefore, we believe the core of the recent tech volatility lies in the current earnings void between China and the US, where markets lack clear fundamental guidance. This forces capital to trade on marginal industry changes, some of which may constitute mere "noise."
Several concerning narratives have contributed to the amplified volatility in global tech stocks recently:
1) The accelerated pace of bond issuance and changes in financing structures by North American cloud providers have raised market concerns about their ability to cover expanding capital expenditures with internal cash flow. Coupled with a recent decline in token usage, leading tech giants, particularly the "Mag7," have led the decline, becoming a significant drag on the Nasdaq.
2) Rising memory chip prices forcing price hikes for products from companies like Apple and Microsoft have heightened fears about the sustainability of massive capital expenditures, should high computing costs stifle end-user demand. This has led to declines for both global "computing users" and "computing producers."
3) OpenAI's announcement of a delayed IPO has sparked worries that weaker-than-expected operational data from a leading star company could prompt more cautious investment across the industry.
These market concerns all revolve around a key metric: the growth rate of Hyperscalers' capital expenditures (Hyperscalers refer to Microsoft, Google, Amazon, Meta, and Oracle). Historical analysis shows this is indeed a simpler, more effective, and forward-looking indicator for tracking and predicting AI industry trends. Whether for US hardware or software stocks, the consensus year-over-year earnings expectations for the next 12 months are strongly correlated with Hyperscalers' year-over-year capex growth.
However, many factors influence capital expenditures, and numerous data points and changes represent "noise." The key is to focus on the primary drivers:
Firstly, at its core, changes in Capex are primarily determined by return on investment (ROI), which is currently linked to the Annual Recurring Revenue (ARR) of leading large language model developers. This data will become more transparent after Anthropic's potential IPO and is expected to continue supporting a "virtuous cycle" of rising ROI and Capex. As long as computing power continues to enhance AI performance and the ROI from token generation remains attractive, leading AI model developers will keep increasing capital investment. Cloud service providers, as "computing suppliers," earn sufficient indirect benefits to support further increases in their capital expenditures.
Secondly, while the market expects the existing free cash flow of North American cloud providers to turn negative this year, questioning capex sustainability, the more critical factor is whether future growth in operating cash flow can support high capex growth. As long as the growth in operating cash flow from cloud and AI businesses outpaces the growth in capital expenditures, free cash flow will eventually turn positive, supporting sustainable high capital spending. So far in 2024, forward revenue expectations for major North American cloud providers have been consistently revised upward. The upcoming earnings season in mid-to-late July will be a crucial window to verify whether their AI-related revenues can justify high capex growth.
Finally, historical patterns suggest earnings seasons are important windows for upward revisions to capex expectations. For 2024-2026, initial annual capex growth expectations at the start of the year have been roughly three times lower than the final figures. Each earnings season, when listed companies provide new guidance, presents a significant opportunity for the market to revise its capex expectations upward.
In summary, the recent volatility in global tech stocks stems partly from structural crowding and the amplifying effect of extreme leverage. However, at its heart, it reflects a tug-of-war among earlier investors over global capex expectations and marginal industry changes, occurring in a vacuum of clear fundamental guidance during an earnings lull. The fundamental picture for the tech industry has not seen substantial deterioration. The upcoming earnings season will determine the true strength of the tech sector's momentum. The recent divergence and volatility caused by certain "noise" may, in fact, create opportunities for strategic positioning.
Domestic Market: The Root of Recent Structural Divergence—Momentum—Faces a Key Validation Window
For the A-share market, beyond global AI industry dynamics, we have previously emphasized that the recent structural divergence is fundamentally driven by domestic momentum differentials. The upcoming pre-earnings announcement period in early July will serve as another validation of this divergence. This means the consensus favoring growth and momentum-driven themes is unlikely to change due to recent volatility.
Currently, data from macro, meso, and micro perspectives all point towards a potential widening of the momentum gap between emerging and traditional industries in the mid-year reports:
1) Macro: May retail sales and fixed asset investment growth turned negative, while growth in AI-related high-tech industries and exports continued to rise.
2) Meso (Industry): AI-related TMT and upstream resource sectors (mainly non-ferrous metals, chemicals) have seen continuous growth in industrial enterprise profits year-to-date, while sectors like consumer goods, midstream manufacturing, and utilities have continued to decline.
3) Micro (Company): Post-Q1 reports, only TMT hardware and upstream resource sectors have seen upward revisions to their 2026 net profit estimates, while other sectors have faced downward revisions.
Therefore, as the market's primary focus in July shifts to industry trends and earnings validation, the consensus favoring growth and momentum-driven themes is not expected to change due to recent fluctuations.
Investment Strategy: Stick to the Momentum Theme, Avoid Unnecessary Rotation
In terms of portfolio allocation, there is no need to rotate for rotation's sake. The relative strength of momentum and changes in earnings performance remain the core guiding principles. Whether looking at changes in consensus earnings estimates post-Q1, the latest industrial enterprise profit data, or institutional investor surveys, consensus on momentum remains highly concentrated.
AI computing hardware (optical communication, semiconductors, PCBs) still represents the strongest consensus. Some advanced manufacturing sectors (AI upstream equipment, batteries/energy storage, shipbuilding) have also seen upward revisions to their momentum expectations. Additionally, upstream resource sectors (many also driven by AI demand, such as minor metals, energy metals, chemicals) are in focus.
Within the AI sector, considering positioning, relative valuations of leaders, and overseas parallels, the recent significant declines in the North American computing chain (optical modules, PCBs) due to overseas influences present more attractive opportunities for positioning during the July earnings validation period, given their underlying momentum advantage. Domestic AI software and services (IT services, cybersecurity, enterprise software) have further room for gains compared to their overseas counterparts.
Furthermore, the recent significant gains in US biotech/pharma stocks (innovative drugs, AI in medicine, brain-computer interfaces), as a sector playing catch-up from lower levels, warrant attention for their potential spillover effects on A-shares.
Risk Factors: Fluctuations in economic data, policy easing falling short of expectations, the Federal Reserve's rate cuts being less aggressive than anticipated, escalation of geopolitical tensions.
Comments