---
title: "A Small City Counts Its Computing Power First"
date: 2026-11-05
category: Companies & Stack
site: NeuroAI
canonical: https://neuroai.site/a/na-jiaxing-compute-26-1105b
language: en
---

# A Small City Counts Its Computing Power First

> Running a household means keeping accounts, and a city that wants to build a new industry has to know what it has in hand first. What Jiaxing counted first was an invisible ledger: computing power.

Running a household means keeping accounts, and a city that wants to build a new industry has to know what it has in hand first. What Jiaxing counted first was an invisible ledger: computing power. It does not sit on a shelf, nor is it written on a sign, yet it decides how much work can be taken on later and how large an operation can be supported.

According to People's Daily, as a supporting event of the World Internet Conference Wuzhen Summit, the 2026 Digital Yangtze River Delta Development Conference was held in Wuzhen, Jiaxing, on November 4, with about 300 guests from government, industry, academia, and research. A set of numbers at the conference laid the city's assets out in the open; this piece takes only the entries that can be checked.

This piece sets out to do one thing: take that set of numbers apart and say which is revenue, which is growth, which is scale, and which is share. They look alike but are not the same thing at all; read them mixed together and mistakes come easily, and once the order slips the whole ledger slips with it.

## Laying Out an Industry Ledger

Start with money. According to People's Daily, in the first three quarters of this year, Jiaxing's core AI industry recorded revenue of 710 hundred million yuan, up 26% year on year. The two numbers are a pair: one is the total revenue accumulated over that stretch, the other is how much higher that is than the same period a year earlier, and neither works alone.

The words "first three quarters" cannot be dropped. They refer to the cumulative figure for January through September 2026, not to a single quarter and not to the whole year. Without that period, the 710 hundred million yuan has nothing to stand on, and readers easily take it for a full-year figure — and every step they reason from there goes wrong.

That People's Daily piece carried a source note at the end. This piece follows its wording and does not extrapolate: it does not guess what the full year might reach, nor does it estimate next year's scale for the city. A cumulative figure is a cumulative figure; writing the period clearly is enough, and one less guess actually puts readers more at ease.

For ordinary people, this ledger turns into one plain sentence: the business tied to artificial intelligence in this city added up to that much over nine months, a notch more than the same period a year earlier. That is all; there is no need to imagine something larger, still less to treat it as a whole year's harvest.

## Revenue and Growth Are a Pair

The 710 hundred million yuan and the 26% have to be read together. The first is an absolute figure, answering how big; the second is a relative one, answering how fast. Pull either one out on its own and it is easy to read a meaning the original never had, and the two no longer add up to a complete statement.

Still less can the 26% be described as "computing power growth." It sits after revenue and speaks to the year-on-year change in industry revenue, which has nothing to do with how many machines were added to the data halls. Move the figure one place and the meaning changes entirely; carry it further and it becomes a different claim.

The report also mentions a change in revenue from computing-power data services, which this piece does not pass along. The reason is the same as before: different indicator, different caliber; set side by side, they are most easily read as the same thing being said twice, which would only drown the main thread.

For a piece written for ordinary readers, it is better to explain one pair of numbers thoroughly than to copy a whole page of percentages. Knowing which two figures belong together and what each one answers is far more useful than memorizing several more, and less likely to be mixed up later.

## Scale and Share Are Not the Same

Now to the machines. According to People's Daily, Jiaxing has put 118 EFLOPS of computing power into operation, accounting for more than 60% of the province's total computing scale. The unit measures how fast calculation runs; ordinary readers need not remember it, only know that it speaks to "how much work can be run," without memorizing that string of letters.

Pause here: the 118 EFLOPS is the scale of computing power already switched on and usable; the more than 60% is that scale's share of the province's total. One is the city's own stock, the other is a proportion measured against elsewhere — they are not the same indicator at all, and placing them together does not mean they should be added up.

So it must never be written as "computing power grew by sixty percent." The original says the share is over half, not that it rose by that much over last year. A few words' difference sends the meaning down another road entirely; this piece sets them apart on purpose so readers do not casually merge them into one sentence.

The report also mentions computing scale still under construction, which this piece does not carry over either. What is already in operation and what is still being built are two separate ledgers and need to be recorded apart; folding the latter into existing capacity inflates the assets and leads readers to overrate what can actually be used today.

## What the Report Says

A document was released at the conference as well. According to China News Service's Zhejiang channel, the Digital Yangtze River Delta Development Report (2026) was formally released there, with someone on site giving a systematic reading of its contents. Writing the name in full is so that anyone who wants to look it up later does not come up empty.

Do not drop the year or the word "Development" in the title. It is the document's formal name; written in full, a reader who wants to check back can find it. Call it casually "a report" or "the Yangtze River Delta report" and it no longer matches anything, leaving even the source with nowhere to be traced.

This piece records only the layer of "what was released and what it is called"; it does not pass along the report's judgments, nor draw conclusions on its behalf. The document is one thing and the reading of it another. Blend the two into a single sentence and the wording easily grows too large, leaving readers unable to tell which line came from the original.

## Twelve Scenarios and One Initiative

Beyond the numbers, several concrete things came out of the conference. According to People's Daily, the event presented 12 "AI+" high-value application scenario pioneer practices from Shanghai, Jiangsu, Zhejiang, and Anhui, and partner cities jointly launched an initiative to build and share scenarios — all of them arrangements laid out at the conference.

The 12 are a batch of selected application examples, not this city's own report card. They come from Shanghai, Jiangsu, Zhejiang, and Anhui, set out so each place can look, compare, and try them in turn; do not read them as one party's submitted answers, still less as a local ranking.

That initiative was summed up as "innovate in one place, reuse in many," meaning the same approach can carry on in another city. According to People's Daily, the conference also arranged a pain-point question-and-answer dialogue session; this piece treats it as background only and does not go into what was said there.

It does not say which model deserves to be spread, nor judge whether these practices are good or bad. Which scenarios can truly be reused and how much must change in a new place can only be known by using them; a piece of writing cannot stand in for that judgment, and such words are better left to those who have actually tried.

## Count First, Then Begin

To gather up that opening set of numbers: first-three-quarter revenue and year-on-year growth are one pair; computing scale already in operation and its share of the province's total are another. The two pairs each speak for themselves, neither vouches for the other, and keeping them apart is what stops them from bleeding into one another.

What is interesting is the order. Rather than first declaring how big it intends to become, the city first counted the machines it already had and the revenue it had already produced. Get the accounts right, then talk about where to go — that is the plainest rule of running a household, and it holds for an industry too.

Said to a relative who does not follow the stock market: it is like meaning to buy a large appliance at home — first check the bankbook and take stock of what is in the cupboard, then decide whether to buy. The technology sounds distant, but this step of counting is no different from anyone's household budgeting, and nobody needs to understand those letters first.

**Disclaimer: This article is based on publicly available information and is for reference only. It does not constitute investment advice.**

---

Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-jiaxing-compute-26-1105b
Free to quote with attribution and a link to the original.
