Various signs indicate that we are "approaching the cycle peak," but UBS insists: it is "too early" to exit Chinese tech stocks now.
```
Just after a sharp drop in global chip stocks, UBS has assessed China’s AI tech market: this rally has indeed sounded an alarm, but it’s not yet time to retreat.
According to ZhuiFeng Trading Desk, on June 23, the UBS Asia Equity Strategy team, including analyst James Wang, stated in a report that there are indeed several signals that China’s AI tech hardware stocks have “approached a cyclical peak,” but profit momentum remains strong, and exiting now would be premature; they maintain an overweight position.
Top signals have emerged, but it’s not the end
The report points out four cautionary signals:
First, positions are highly crowded. The proportion of domestic public funds’ holdings in A-share telecom and electronics sectors has reached near historical highs, and trading concentration data also show apparent herding.
Second, valuations are no longer cheap. The forward P/E ratio of China’s AI tech hardware stocks is close to historical highs, and more importantly, compared to global peers, their valuation discount has disappeared—which has not been the norm in recent years.
Third, a higher number of IPOs. The supply of new shares is accelerating, which historically often coincides with market peaks.
Fourth, capital expenditures are continuously rising. However, this round of production expansion is relatively restrained, partly due to limited supply of high-end manufacturing equipment.
These four signals together outline a basic picture of being “near the top.”
But rapid profit growth is the trump card
Why not exit yet? The answer lies in earnings data.
Currently, the expected earnings growth rate for China’s AI tech hardware stocks is around 80%. Historical data shows that when earnings growth exceeds 30%, tech stocks typically outperform the broad market by 10% to 20% per year. The earnings growth outlook for 2027 is also expected to remain above 30%.
In the past three months, earnings expectations for China’s AI tech hardware stocks have been revised up by an average of 15%, while overall market earnings expectations were revised down by 2% over the same period. This divergence is the fundamental support behind the crowded positions.
Contract liabilities and inventory as a percentage of revenue continue to rise—these two indicators are leading signs of future order visibility, and have risen significantly since the release of DeepSeek (early 2025). Order visibility now extends to the end of 2027.
History shows: The last phase is often the wildest
Reviewing the 4G, 5G and cloud computing tech cycles, each lasted about 2 years, with excess returns of about 100% for related stocks, and P/E expansion exceeding 19x.
This round of the AI rally, which started since the DeepSeek moment in early 2025, has already achieved a 215% excess return and a 16x P/E re-rating—by the numbers alone, it has far exceeded historical averages.
But here’s a detail worth noting: Historically, in the last three months before tech stocks peak, related sectors averaged another 48% gain. In other words, even “near the top,” the final leg may not be short.
Of course, risks are real. Historical data shows that after revenue growth peaks, related AI tech stocks underperform the market by about 2% on average over the following six months. While not disastrous, this means greater caution is needed at the inflection point.
The demand chain is still intact: order visibility extends to end-2027
From an industry chain perspective, AI data centers don’t buy only GPUs.
More powerful chips lead to greater power loads, more complex cooling, faster data transfer, higher-end packaging, higher-spec PCB, and upstream material requirements. AI hardware demand spreads beyond just the chips themselves.
UBS’s industry framework points to several key observations:
The entry barrier for high-end tech components is rising “exponentially,” which may widen the gap between leaders and non-leaders.
Demand for AI tech components remains strong, with order visibility through the end of 2027.
Chinese tech companies are catching up in AI capabilities; some also benefit from high-end AI product demand crowding out low-end consumer electronics component production capacity.
Capital expenditure is rising, but this round of expansion is relatively restrained, partly due to limited high-end manufacturing equipment supply.
AI data center construction is still on an upward trend, with the US building faster than China.
In this round of Korean chip stock volatility, HBM is a key sensitive term. UBS notes that HBM has shifted from a supporting component to the central bottleneck in AI hardware deployment, as large-model training and inference need higher memory bandwidth. But China currently lacks direct HBM suppliers and is instead an indirect beneficiary link.
Compared to global peers, “cheap” is no longer China AI hardware’s only story
Both Chinese and overseas AI hardware stocks have seen their fundamentals expand markedly in recent years.
Since 2021, revenue for both China and global AI tech hardware companies has roughly doubled. But since then, global peers have outpaced Chinese ones in earnings growth, free cash flow scale, and margin levels.
In detail:
- Stock price performance: In recent years, global AI tech hardware stocks cumulatively outperformed Chinese peers by about 130%; but in 2026, Chinese stocks overtook, outperforming by 17%.
- Valuation: Chinese stocks still lag global peers in P/B and P/S, but are no longer cheap on a P/E basis.
- Profit quality: Global peers’ net profit margin is at historic highs, Chinese peers have only slightly improved; global peers’ free cash flow scale is far higher; global peers are also more attractive on PEG, after adjusting for growth.
- Growth pace: Global peers’ profit growth will be faster in 2026, China will be slightly faster in 2027—this is a relative advantage window for Chinese stocks.
2027 will be bumpier, leadership differentiation is the main theme
Analysts remain positive for the rest of this year, as revenue and profit growth remain strong.
But 2027 may prove more difficult. As capacity expands and the growth base rises, the market may become more turbulent. The report says, “2027 will be a key year to test whether these companies and industries can truly deliver on their AI promises.”
If the rally enters its latter phase, small-cap tech companies may benefit from sector beta and spillover demand. But in the long-term, those segment leaders able to win global supply chain orders, with pricing power and tech advantages, will more easily win out.
UBS prefers segment leaders in optical modules, memory, GPU, copper-clad laminate, and semiconductor equipment.
2027: The real test is only just beginning
Analysts’ final judgment: for the rest of 2026, strong revenue and profit growth will continue to support AI tech stocks. But in 2027, as capacity expands and high-base effect appears, the market may become even bumpier.
Companies that can truly deliver on the promise of AI and keep entering the global supply chain will be the winners in the next phase. Those that just rode the tailwinds of this round may face harsher scrutiny then.
In the original words: 2027 will be “a key year to test which companies (and industries) can truly deliver on the promise of AI.”
~~~~~~~~~~~~~~~~~~~~~~~~
The above highlights are from ZhuiFeng Trading Desk.
For more detailed interpretation, including real-time analysis and frontline research, please join [ZhuiFeng Trading Desk•Annual Membership]
Risk disclosure and disclaimerMarkets contain risk, investment must be prudent. This article does not constitute personal investment advice, nor does it take into account individual users’ particular investment goals, financial status, or needs. Users should consider whether any opinion, viewpoint, or conclusion in this article matches their specific circumstances. Investing based on this information is at your own risk. ```