UBS: Deleveraging in China's AI Tech Stocks Nearing End, Valuations Retreating to Historical Averages
Nashnova编辑部
UBS says China AI tech hardware stocks fell 32% on average in July while margin balances dropped to RMB 2.6 trillion, and sees the worst of the technical selloff as over — calling this a re-entry window.
How bad was the selloff?
UBS-tracked AI tech hardware stocks fell 32% on average in July; 36% of names dropped 40% or more within the month.
A-share margin balances retreated from a RMB 3 trillion peak to RMB 2.6 trillion, near pre-April leverage levels.
This means → most of the leveraged money has been flushed out; the most violent phase of technical selling is likely behind us.
Where do valuations sit after the drop?
AI tech hardware valuations have pulled back sharply, now only slightly above historical averages.
Meanwhile, earnings-per-share forecasts are still being revised upward — prices fell, but profit expectations did not.
In plain terms = cheaper price, intact fundamentals — that is UBS's case for re-entry.
What changed in the global AI narrative?
UBS cites four positives: Microsoft and Amazon results showing strong OpenAI growth and improving AI monetization; robust demand from AI-native companies; rising backlog at major cloud providers; enterprise AI spending up 25% quarter-on-quarter.
China AI tech hardware was dragged down by the global selloff, but the domestic supply-chain narrative and fundamentals have not changed — domestic GPU supply is improving, and AI data-center buildout is expected to accelerate in H2.
This means → as global AI confidence recovers, investors have reason to refocus on China AI supply-chain fundamentals.
Is there still a safety cushion?
The overall A-share collateral ratio stands at roughly 280% of margin loans — enough to absorb another round of selling.
Hong Kong government support for tech ETFs also helps anchor market confidence.
In plain terms = leverage is down, collateral is thick, policy backstops are in place — three buffers stacked together.
What does UBS favor for H2?
Within AI hardware, the preference goes to semiconductor equipment, networking chips, and advanced packaging — the less cyclical links.
Beyond hardware, UBS likes four themes: internet (improving earnings + cheap valuations), power equipment (data-center buildout + energy self-sufficiency), non-ferrous metals (strong earnings trend), and "going global" stocks (fading RMB appreciation drag).
This reflects UBS's view that H2 market breadth will widen — capital will no longer funnel into AI hardware alone.
What are the risks of re-entering now?
UBS warns: recent price swings have been sharp, and market concerns about AI monetization have not fully faded.
Investors may not embrace AI hardware as eagerly as in H1; stock performance could become far more dispersed.
This means → the key test for H2 is whether previously pressured sectors like internet can actually absorb the capital flowing out of AI hardware.
Content is for reference only, not financial advice.