Shenwan Hongyuan: AI Contributes Nearly 2.3 Percentage Points to PPI, Double-Peak Pattern Likely for the Full Year
N.R. Finch
Shenwan Hongyuan estimates AI-linked inflation contributes nearly 2.3 percentage points to current year-on-year PPI, and combined with a second oil-price surge and petrochemical supply exits, full-year PPI may trace a double peak rather than the single-hump decline the market expects.
Why is the market's "single-peak inflation decline" consensus wrong?
The mainstream view assumed PPI would peak once and slope down. Shenwan Hongyuan identifies three under-appreciated forces: a second oil-price surge, accelerating petrochemical supply exits, and structural price increases driven by AI.
This means → full-year PPI draws not one hill but two peaks (a double top), while CPI may produce three interim highs (an M-shaped path).
In plain terms = inflation will not simply "crest and coast downhill" — it climbs again midway.
Why did oil prices swing from drop to surge, and how does that hit PPI?
A US-Iran memorandum of understanding in mid-June briefly pushed oil to around $70/barrel, but the MOU has no legal force — inspection rights and enriched-uranium disposal remain unresolved.
The Strait of Hormuz has since been disrupted again, sending oil back above $90/barrel. EIA crude inventories have fallen from roughly 850 million barrels in early May to 720 million barrels, near historic lows.
This means → international oil prices lead China's petroleum-extraction PPI by about two weeks: June's roughly 20% oil plunge will drag July PPI, while the mid-July rebound will lift August PPI — creating a "July low, August high" rhythm.
Base case: year-on-year PPI dips to 3.6% in July, then re-climbs to 4.1%–4.2% in August–September, with the second peak likely in September before easing back to 3.6% by year-end.
What does petrochemical "supply exit" mean, and why does it keep inflation sticky?
Supply exit — weaker firms shutting down and capacity shrinking — has been severe: petrochemical utilisation rates have dropped to roughly 22%, a record low, employment growth has fallen to about -2%, and the number of firms has turned negative for the first time.
In plain terms = the more factories close, the greater pricing power the survivors hold — even when crude drops, midstream and downstream product prices barely follow.
The data confirm this: in June, petroleum-extraction PPI fell 18.9 percentage points year-on-year, while midstream and downstream petrochemical PPI fell only 0.9 percentage points — a stark upstream-downstream divergence.
This reflects a pattern seen after the 2017 and 2021–2022 oil rallies, where upstream PPI fell with oil but midstream and downstream PPI stayed resilient for six months to a year, across chemicals, chemical fibres, rubber, plastics, and textiles.
How has AI become an inflationary force too?
Direct channel: PPI for electronic components and computer-communications equipment is rising fast — AI computing demand is pulling hardware prices higher.
Indirect channel: AI investment lifts non-ferrous metal prices (e.g. copper), which feed through to metal products and equipment PPI.
Together the two channels contribute nearly 2.3 percentage points to current year-on-year PPI and roughly 0.2 percentage points to CPI (mainly via memory-chip price increases feeding into smartphones and other consumer electronics).
In plain terms = AI is not just reshaping the tech sector — it raises the cost of chipmaking materials, and ultimately you pay a share of the AI boom every time you buy a phone or an appliance.
How does producer-price inflation reach consumers step by step?
The transmission path runs: "producer-goods PPI → consumer-goods PPI → core goods CPI," with a lag of roughly four months.
Since January, fast-rising producer-goods PPI has already lifted non-food consumer-goods PPI from -1.5% in March to -0.1% in June; core goods CPI (excluding gold and jewellery) moved from 0% to 0.6% over the same period.
This means → the upstream price signal is working its way down the supply chain, and consumer-level price pressure has not yet peaked.
What shape will full-year CPI take?
The report forecasts an M-shaped "triple-peak" CPI path: the three highs are February (1.3%), April–May (1.2%), and September (1.0%).
July CPI may slip to around 0.5% on the back of June's oil-price drop, but August–September should climb again as oil rebounds.
Downside constraints: end-consumer demand is recovering slowly, Q2 hog inventory growth remains elevated (limiting pork-price gains in H2), and a high gold-price base will also weigh on CPI readings.
This reflects the two key variables that will test the "double-peak" thesis: whether AI inflation continues to provide structural PPI support, and how long the petrochemical supply exit lasts.
Content is for reference only, not financial advice.