China’s AI compute surge heavily reliant on US chips, offshore cloud access: report

Read the original at South China Morning Post ↗
Cite

Liu, H. (2026, October 24). China’s AI compute surge heavily reliant on US chips, offshore cloud access: report. South China Morning Post. https://www.scmp.com/tech/tech-war/article/3371929/chinas-ai-compute-surge-heavily-reliant-us-chips-offshore-cloud-access-report

Dated from the outlet's own feed. Check the original for its byline.

South China Morning Post · collected 2026-10-24 · by Howard Liu

Quick Summary

A new report from the Institute for Progress indicates that China’s artificial intelligence computing capacity will double to about 8.8 million H100-equivalents by 2027, with US-designed chips accounting for approximately 65% of this capacity. Despite a significant reliance on American technology, Chinese AI models are still advancing rapidly and lag behind their U.S. counterparts by only six and a half months. The report also highlights that remote access to offshore cloud services will see a substantial increase, reaching 2.8 million H100-equivalents by the end of 2027.
Written locally by qwen2.5:14b on 2026-10-24, using this article's own text rather than the other coverage of the same event (that is the story summary below).

AI analysis runs on qwen2.5:14b, locally

Story summary

According to a recent report by the Washington-based Institute for Progress (IFP), China's access to artificial intelligence computing power is set to double next year from about 4.6 million H100-equivalents this year to 8.8 million in 2027, with US-designed chips accounting for about 65% of that capacity. Despite this growth, the US still maintains a roughly 16-fold advantage in AI training compute over China. However, leading Chinese models lag their American counterparts by only 6.5 months, indicating rapid progress. The report also highlights China's heavy reliance on offshore cloud access and US chips for its AI compute surge, with remote access alone expected to rise from around 1 million H100-equivalents this year to 2.8 million by the end of 2027.

Written for “China's AI Compute Reliance On US Chips” on 2026-10-24, grounded in this article and the 0 other(s) covering the same event.
Why this leaning score
This article does not take a side on a contested political question, so it has no leaning score. That is an answer rather than a gap: a match report or a rescue can be warmly or critically written without being left or right, and scoring it anyway is how approval of a subject gets recorded as a political position.
No political leaning scored for article 103283 · logged 2026-10-24

Story

📰 China's AI Compute Reliance On US Chips
Technology · 1 article(s) covering the same event.

Compared with similar articles

This article reads unscored and hedges 50% of its claims. Each row says how that neighbour differs.
China's economic growth picks up as AI boom boosts exports different event · 95%
BBC News
⚖️ Centre 🔴 0% hedged 0 of 12 📰 publisher trust 69
“The articles discuss different aspects of China's AI industry and economic growth, not the same specific incident.”
US-China AI talks must ensure safety of both tech and business different event · 95%
South China Morning Post
⚖️ Leans right 🔴 33% hedged 3 of 9 📰 publisher trust 64
“The articles discuss different aspects of AI development and cooperation between US and China, but do not describe the same specific meeting or incident.”
Z.ai founders worked on research funded by China institute that Britain tied to spies different event · 95%
The Straits Times
⚖️ leaning not scored 🔴 25% hedged 14 of 55 📰 publisher trust 68
“Article A discusses Z.ai founders' involvement with research funded by a Chinese institute tied to spies, while Article B talks about China's AI computing power reliance on US chips and offshore cloud access.”
US orders DeepSeek removed from federal systems, citing national security risks different event · 95%
South China Morning Post
⚖️ Leans right 🔴 33% hedged 1 of 3 📰 publisher trust 64
“Article A discusses the US government's directive to remove DeepSeek products from federal systems, while Article B covers China's reliance on US chips and offshore cloud access for AI computing power.”
US-China AI talks must ensure safety of both tech and business different event · 95%
South China Morning Post
⚖️ Leans left 🔴 33% hedged 3 of 9 📰 publisher trust 64
“The articles discuss related topics regarding US-China AI collaboration but describe different aspects and events.”
South Korean chip makers upgrade manufacturing in China different event · 95%
Semafor
⚖️ leaning not scored 🔴 40% hedged 2 of 5 📰 publisher trust 60
“The articles discuss different aspects of China's technological and economic relationship with South Korea and the US, focusing on chip manufacturing upgrades versus AI compute reliance.”
Anthropic plans $750 billion compute spend different event · 95%
Semafor
⚖️ leaning not scored 🔴 100% hedged 4 of 4 📰 publisher trust 60
“The articles discuss different aspects of AI compute capacity and investment, with Article A focusing on Anthropic's plans and Article B on China's reliance on US chips.”
AI is advancing rapidly. Who's keeping it in check? different event · 95%
CBS News
⚖️ Centre 🔴 0% hedged 0 of 2 📰 publisher trust 66
“The articles discuss different aspects of AI development and regulation, with Article A focusing on tech leaders' plans to manage AI risks, while Article B focuses on China's reliance on US technology for its AI compute surge.”
Asia-Pacific AI agent market to surge 14-fold to US$11.2 billion by 2030, Omdia says different event · 95%
South China Morning Post
⚖️ leaning not scored 🔴 33% hedged 1 of 3 📰 publisher trust 64
“The articles discuss different aspects of AI market growth in China and the Asia-Pacific region but describe distinct events with separate forecasts and focuses.”
DeepSeek, MiniMax secure open-source licenses days after new China export rules: state media different event · 95%
South China Morning Post
⚖️ Centre 🔴 33% hedged 1 of 3 📰 publisher trust 64
“The articles discuss different aspects of China's AI industry without describing the same specific incident.”
How this article was measured signals, leaning charts, publisher record, byline, topics and claims

Signals How these are calculated →

Claims extracted
2
sentences that state something as happening or true
Uncertain
50%
1 of 2 hedged ("reportedly", "may", "according to")
Leaning
not political
takes no side on a contested political question
Publisher signals
64/100
South China Morning Post's corrections and hedging, across everything collected from it. Not a measure of accuracy.
Outlets on this story
1
Technology
Same framing elsewhere
1
articles that frame the event the way this one does
Analyzed 2026-10-24 · how these are computed

How this is being covered How these are calculated →

Article leaning vs. publisher reliability
Source leaning vs. consistency

Publisher

South China Morning Post · 2957 article(s) · 6 correction(s) detected
Running correction rate · 6 correction(s)
2026-10-16
Nobel Peace Prize: the message behind Navi Pillay’s win
2026-10-16
Indonesian governor’s Singapore trip adds fury to forest fire, haze crisis
2026-10-04
Flight to Okinawa diverts back to Hong Kong after suspected landing gear fault
2026-10-04
Tennessee prisons chief resigning after botched execution of Christa Pike
2026-09-29
Indonesian prison chief suspended over secret VIP compound for inmates
2026-09-28
Trump ‘arms offer’ to China jolts Washington into moves to contain the fallout

Who wrote this

Howard Liu
10 article(s) here · 1 carrying a prediction
🔮 China’s remote access alone is expected to swell to 2.8 million H100-equivalents by the end of 2027, up from roughly 1 million this year China’s access to artificial intelligence computing power will double next year, but US-designed chips will account for about 65 per cent of that capacity, an indication of its continued reliance on foreign hardware, according to new research. Total compute available to Chinese entities was expected to expand in 2027 from about 4.6 million to 8.8 million “H100-equivalents” – a metric benchmarked against Nvidia’s data-centre graphics processing units, according to a report by the Washington-based Institute for Progress (IFP).
2026-10-24 · mixed framing · China’s AI compute surge heavily reliant on US chips, offshore cloud access: report
🔮 Asia-Pacific AI agent market to surge 14-fold to US$11.2 billion by 2030, Omdia says In the period to 2030, the agentic share of the region’s AI software revenue is projected to rise to 12.3 per cent from 1.7 per cent The artificial intelligence agent software market across Asia, Australia and New Zealand is set to expand more than 14-fold by 2030, according to research firm Omdia, as Chinese tech companies increasingly deploy automated tools across office and industrial operations. During its annual event in Beijing on Thursday, Omdia forecast that regional market revenue, which included mainland China and Hong Kong, would reach US$11.2 billion in 2030, up from an estimated US$782 million this year.
2026-10-23 · mixed framing · Asia-Pacific AI agent market to surge 14-fold to US$11.2 billion by 2030, Omdia says
🔮 TSMC confirms talks with Elon Musk over Texas ‘Terafab’ but insists on operational control The negotiations underscore TSMC’s push to expand production and satisfy insatiable demand for AI chips from key clients like Nvidia Taiwan Semiconductor Manufacturing Co (TSMC) is in discussions regarding a potential role in Elon Musk’s proposed Terafab project in Texas, according to company chairman and CEO C.C. Wei, although he stressed that TSMC would retain operational control of any partner factory. The plant “has got to be run by TSMC”, Wei said during the third-quarter earnings call on Thursday.
2026-10-15 · mixed framing · TSMC confirms talks with Elon Musk over Texas ‘Terafab’ but insists on operational control
🔮 Chinese memory giant YMTC to grow revenue 6 times faster than peers as IPO nears: report Country’s top NAND maker bets on enterprise SSDs and AI infrastructure to drive next phase of growth, Counterpoint says Yangtze Memory Technologies Corporation (YMTC), China’s leading flash-memory maker, is set to grow revenue six times faster than its industry peers this year as it expands into server-related business ahead of a planned initial public offering (IPO), according to Counterpoint Research. The company’s server-related revenue was effectively zero in 2024, but could account for nearly 39 per cent of its revenue by 2028, according to the market intelligence firm on Tuesday.
2026-10-14 · speculative framing · Chinese memory giant YMTC to grow revenue 6 times faster than peers as IPO nears: report
🔮 As it begins massive output increase, China’s top memory supplier sets ‘aggressive’ target for local chipmaking equipment, source says The firm, listed on the Shanghai Stock Exchange’s Star Market, would spend 24.1 billion yuan on a new technology research and development project and another 10.8 billion yuan on the second phase of a wafer testing base, according to a Monday stock filing.
2026-09-29 · mixed framing · China’s CXMT to favour domestic suppliers in US$5.2b memory-chip capacity push: source
🔮 The sum, which Shanghai-listed Biwin Storage Technology will pour into the third phase of its advanced packaging and testing project in Dongguan, in southern China’s Guangdong province, will exceed the 3.09 billion yuan it put into the first two phases combined.
2026-09-23 · mixed framing · China’s storage makers aim to climb value chain with advanced packaging bets
🔮 Capital expenditure by China’s major tech companies was set to more than double to around US$140 billion this year – up from US$65 billion in 2025 – and reach US$165 billion by 2027, according to Moody’s.
2026-09-06 · mixed framing · Punching above their weight: how China’s AI giants stretch each dollar in compute race
Also by Howard Liu
Asia-Pacific AI agent market to surge 14-fold to US$11.2 billion by 2030, Omdia says
2026-10-23 · South China Morning Post
TSMC confirms talks with Elon Musk over Texas ‘Terafab’ but insists on operational control
2026-10-15 · South China Morning Post
Chinese memory giant YMTC to grow revenue 6 times faster than peers as IPO nears: report
2026-10-14 · South China Morning Post
Mapping US chip supply chain: China accounts for far more than it appears, report says
2026-10-12 · South China Morning Post
Nothing else under this byline is closely related to this article, so these are simply their most recent.
All 10 articles by Howard Liu →

Topics

China Chinese Institute for Progress Nvidia Washington

Subjects

China GPE · 4× Chinese NORP · 2× American NORP · 1× IFP ORG · 1× Institute for Progress ORG · 1× Nvidia ORG · 1× Washington GPE · 1×

Narrative

While the US maintains a roughly 16-fold advantage in AI training compute over China, leading Chinese models lag their American rivals by only 6.5 months, the report said.
framing: mixed · carried by 1 article(s) · first seen 2026-10-24
🔮 China’s remote access alone is expected to swell to 2.8 million H100-equivalents by the end of 2027, up from roughly 1 million this year China’s access to artificial intelligence computing power will double next year, but US-designed chips will account for about 65 per cent of that capacity, an indication of its continued reliance on foreign hardware, according to new research. Total compute available to Chinese entities was expected to expand in 2027 from about 4.6 million to 8.8 million “H100-equivalents” – a metric benchmarked against Nvidia’s data-centre graphics processing units, according to a report by the Washington-based Institute for Progress (IFP).
2026-10-24 · South China Morning Post
China’s AI compute surge heavily reliant on US chips, offshore cloud access: report · mixed framing

Claims (2 extracted, 1 hedged)

China’s remote access alone is expected to swell to 2.8 million H100-equivalents by the end of 2027, up from roughly 1 million this year China’s access to artificial intelligence computing power will double next year, but US-designed chips will account for about 65 per cent of that capacity, an indication of its continued reliance on foreign hardware, according to new research. Total compute available to Chinese entities was expected to expand in 2027 from about 4.6 million to 8.8 million “H100-equivalents” – a metric benchmarked against Nvidia’s data-centre graphics processing units, according to a report by the Washington-based Institute for Progress (IFP). uncertain
compute → expect → Progress
While the US maintains a roughly 16-fold advantage in AI training compute over China, leading Chinese models lag their American rivals by only 6.5 months, the report said. asserted
report → maintain → months
💬Give feedback
🕘History 🎫Support
Terms · Privacy · beta v2.0.0
Articles to 24 Oct 2026