苏州18年前投资的三千万，长成了AI的血管

Original Author: Sleepy, Daria

Original Editor: Cozzy

Suzhou's most profitable investment occurred amidst the global financial crisis.

In the fall of 2008, Lehman Brothers collapsed, and risk capital across China ground to a halt. Investors were cutting losses or sitting on the sidelines, and the primary market had frozen over. It was in this window that Yuanhe Holding, the state-owned investment platform of the Suzhou Industrial Park, injected approximately 30 million RMB into a startup.

The founder was Liu Sheng, a Tsinghua undergraduate and Georgia Tech PhD. He had spent years in North America working on optical communications, first at Lucent and then at Opnext. In 2008, he returned from Silicon Valley, registered InnoLight Technology in Suzhou, and aimed to produce high-speed optical transceivers.

In the 2008 investment landscape, optical transceivers were not just niche; almost no investor was looking at this sector. Domestic optical communication companies were all clustered in the telecom market and low-to-mid-end manufacturing. According to a LightCounting report, only one Chinese company, WTD (later merged into Accelink), made the global top ten in 2010, barely scraping in at the bottom. At that time, the judgment that "cloud computing data centers will require massive volumes of high-speed optical transceivers" was light-years away from any consensus.

Eighteen years later, this investment grew into Zhongji InnoLight. Before the Dragon Boat Festival in June 2026, its stock price surged to 1367.88 RMB, giving it a total market capitalization of 1.5 trillion RMB, placing it in the top ten among A-share companies, surpassing Kweichow Moutai.

The Blood Vessels of AI

The optical transceiver module is crucial to the entire AI industry. Without it, AI is just a pile of silicon chips that generate heat. When signals leave the chip and need to travel between data centers via fiber optics, they rely on optical transceivers.

Computing power is the heart; optical transceivers are the blood vessels.

In the A-share market, Eoptolink, Zhongji InnoLight, and TFC Communication are called the "Yi-Zhong-Tian" (a play on a famous historical figure). These three optical transceiver giants control the most critical segment of the global AI computing power supply chain. Among them, two are based in Suzhou: Zhongji InnoLight’s Suzhou operational headquarters is in the Suzhou Industrial Park, and TFC Communication’s operational headquarters is also there.

Beyond these, Suzhou-based Lynxi Technologies, which makes high-end optical communication testing instruments, saw its stock price increase over twenty-fold from its IPO price within two months of listing in 2026, claiming the title of A-share stock king. Yuanjie Technology, a leading active optical chip company and part of Zhongji InnoLight’s supply chain investment, saw its stock price touch 1712 RMB before the Dragon Boat Festival, nearly quadrupling in six months.

Dissecting the AI data center reveals that Suzhou didn't just bet on a few companies; it bet on a series of interconnected positions.

Further upstream are optical chip companies like Yuanjie Technology. Downstream from them is the optical component field where TFC Communication specializes. Zhongji InnoLight packages these components into high-speed optical transceivers that are shipped to data centers worldwide. Lynxi Technologies then handles the testing to ensure these high-speed optical communication devices can run stably.

According to Xinhua Daily, the total market value of A-share listed companies in Suzhou has exceeded 4 trillion RMB, ranking fourth nationally, after Beijing, Shanghai, and Shenzhen.

Hangzhou's "Six Little Dragons" made it to the Spring Festival Gala and trended on social media. Shenzhen's humanoid robots were dancing. Three GPU companies in Shanghai were telling grand stories of domestic substitution. These stories are flashy, novel, and visual – a 15-second short video is enough to etch a name into memory.

But Suzhou tells a different kind of story.

A fiber optic cable manufacturer in Wujiang District saw its 2026 sales revenue grow over 35% year-on-year, with production capacity already booked through 2027. Multiple Suzhou-based industry chain companies jointly invested to start construction of the nation's first 8-inch silicon photonics chip mass production line. Suzhou's optical communication industry has coalesced into a hundred-billion-yuan cluster, with companies taking the lead in forming over 50 innovation consortia. The industry's scale grew by over 60% in the first quarter of 2026.

When this AI computing boom erupted, real money first flowed to the places building the "blood vessels."

For Suzhou, when a leader in optical transceivers grows, it brings more than just market capitalization. It brings along its suppliers, testing equipment makers, optical chip companies, funds, and talent. Its orders become the capacity for upstream and downstream companies. Its IPO becomes the credit for the local capital market. Its judgment on the industry, in turn, becomes clues for Suzhou's next round of investment attraction and industrial strategy.

Beyond the paper gains of investment returns, what Suzhou has earned is an industry chain that can continue to roll forward.

In 2024, the world suddenly discovered optical transceivers and rushed into the secondary market to grab shares. But Suzhou's name was already inscribed in InnoLight’s earliest business registration records, as an old shareholder from 2008.

Squatting on the Ground to Pick Projects

Why did Suzhou make this successful bet?

Some call it foresight, astute vision, farsightedness. But these are all 20/20 hindsight remarks.

You need to stand tall to have vision, but Suzhou wasn't standing tall. It was squatting on the ground, examining projects one by one, conducting interviews with people one by one, and then making judgments.

The person making the decision for the city was Dai Yu, the project lead at Yuanhe Holding responsible for InnoLight. Years later, in a media interview, she recalled that the initial contact with InnoLight came through the park's "Leading Talent Program." A large number of high-end talents returned to China with projects, and Yuanhe screened them one by one before choosing to invest in InnoLight.

Yuanhe later continued to invest round after round through three different funds at various stages, plus a debt platform, and even brought in Daita Capital from Dongshahu Fund Town to co-invest. Restructuring documents show that the Yuanhe system held over 7% of InnoLight's equity cumulatively.

From investing on the eve of the 2008 financial crisis to InnoLight's backdoor listing via Zhongji Equipment in 2017, Yuanhe traversed an entire economic cycle.

Around 2011, InnoLight's 40G products passed Google's certification, leading to contacts with Amazon and Huawei. In 2014, Google Capital led a $38 million Series C round, its first investment in China. InnoLight originally planned to list on the US stock market, but in 2015, when Chinese concept stocks faced a valuation winter, it turned back, dismantled its variable interest entity (VIE) structure, and sought an A-share IPO.

In 2017, Zhongji Equipment acquired 100% of InnoLight for 2.8 billion RMB, completing the backdoor listing. The de facto controller of Zhongji Equipment, Wang Weixiu, made the bold decision to delegate significant authority to the InnoLight team. New Fortune Magazine used the term "big picture thinking" to describe this decision. By 2026, the wealth of Wang Weixiu and his son Wang Xiaodong had grown to 97.8 billion RMB, their ranking soaring from 161st to 30th. The owner of a small company making motor winding equipment, by acquiring a Suzhou-based optical transceiver startup and being willing to delegate, saw his wealth approach 100 billion RMB.

In 2025, Zhongji InnoLight reported revenue of 38.2 billion RMB and net profit of 10.8 billion RMB, both doubling. In Q1 2026 alone, quarterly revenue reached 19.5 billion RMB, a 192% year-on-year increase, with net profit of 5.7 billion RMB, a 262% increase. The sapling had grown into a forest. InnoLight, through its subsidiaries, invested in a string of industry chain companies like Jingyan Intelligent, Aoke Optoelectronics, and Leishen Photonics, and also invested in Yuanjie Technology, the new A-share stock king.

Looking back at Suzhou's investment track record over the past two decades, InnoLight is not an isolated case. In 2006, "innovative drugs" were not yet a keyword in the capital market. Suzhou established BioBAY, which today houses hundreds of pharmaceutical companies and has produced over thirty listed firms. In the same year, when nanotechnology was far from industrialization, Suzhou began attracting national-level nanotech research institutions. In 2008, its investment promotion team traveled to Cambridge, UK, and brought back the voice AI startup Speechocean. It was still a full nine years before the national "New Generation Artificial Intelligence Development Plan" was issued.

Biomedicine, nanotechnology, voice AI, optical transceivers. Four sectors, all of which Suzhou bet on when others couldn't yet see the potential.

Underpinning this approach is a system that is perhaps not very glamorous but is exceptionally solid. In 2006, the Suzhou Industrial Park established the nation's first dedicated Technology Investment Promotion Center. By 2022, all personnel in the investment promotion line held master's degrees or higher, mostly in science or engineering. Suzhou currently has 212 investment institutions and over 2,200 full-time investment promotion professionals. Among them, 119 are state-owned companies, free from administrative headcount restrictions, recruiting with market-based salaries, and able to directly invest in companies. They are precisely doing what the government wants to do but cannot do through administrative means.

In terms of investment judgment, Suzhou also innovated with the "Chain Leader Fund," where the industry chain leader acts as the fund manager, and government funds co-invest. For example, in the biomedical chain leader fund, Innovent Biologics and Jiangsu Alphamab directly serve as managers, involved in project selection and post-investment management. Over ten such chain leader funds have invested more than 20 billion RMB cumulatively.

So Suzhou's successful bets are not due to luck; it has transformed "gambling" into a robust system. Screen projects with people who know the industry, make judgments with help from veteran supply chain players, and place bets with patient capital over a decade or more.

This is more than just foresight.

Three Generations

This spirit of grinding it out was born out of poverty.

At the start of the reform and opening-up, Suzhou's industrial base was very weak. Equipment was outdated, the industrial structure was backward, and it was disconnected from both raw materials and markets. So, it had to do whatever it could and learn from whoever it could.

The "Shanghai Sunday Engineer" phenomenon was a product of that era. Township enterprises in Suzhou would invite engineers from large state-owned factories in Shanghai during their weekends to guide production. By 1988, there were over 15,000 such township enterprises. By 1992, their output value accounted for over 70% of the city's total.

But even achieving this, when the central government designated Special Economic Zones and Economic Development Zones, Suzhou didn't get a single one.

Policy dividends were unattainable.

Kunshan City simply decided not to wait. They pooled 500,000 RMB of their own funds and established the nation's first self-funded development zone, bearing all the risks themselves, and ran advertisements in Shanghai to attract investment. By adopting the sober and somewhat humble positioning of "doing what Shanghai doesn't want to do" and "Shanghai handles zero-to-one, Suzhou handles one-to-ten," Suzhou wedged itself into the Shanghai-dominated industrial chain, completing its initial industrial accumulation.

Later, Suzhou learned "door-to-door investment promotion" from Singapore, setting up investment offices globally and proactively contacting companies one by one. By 2012, Suzhou's actual utilized foreign capital accounted for 8.1% of the national total, a 131-fold increase from 1990.

Later still, production costs rose, environmental capacity was maxed out, and labor-intensive industries began to move away. Again, Suzhou had no choice but to face the pains of transformation head-on.

Within a 25-square-kilometer radius around Nanjing Road in the Taicang High-tech Zone, there are over 800 small and medium-sized foreign-invested enterprises, forming a complete automotive supply chain where 70% of the components for a complete vehicle can be sourced. In the Silk Road Nanhua Cross-border E-commerce Industrial Park in Shengze Town, an 86-story building houses 86 textile companies, from raw materials and fabrics to finished products and sales, where moving up or down a floor connects you to the next link in the value chain.

These industrial clusters were not planned; they grew organically, one company at a time, one order at a time, over several decades.

In the 1980s, they secretly learned from Shanghai engineers on weekends. In the 1990s, they gambled 500,000 RMB of their own money on a development zone. In 2008, they gambled 30 million RMB on a niche, overseas returnee working on optical modules.

When no one saw potential, they used their own money to buy something that was not yet valuable. Three generations repeated the same action three times.

Visionary people choose their battleground from a height. People with no choice squat on the ground, pick up what others discard, polish it, and wait for it to become valuable. Suzhou's pragmatism is the survival strategy of a prefecture-level city that didn't get a special zone license. It used this strategy for thirty years, and it became part of its bone marrow. When Dai Yu faced InnoLight's business plan in 2008, the neural circuit she used to make her judgment was the same one the people of Kunshan used thirty years earlier when they pooled their 500,000 RMB.

This thing isn't valuable now. But it's worth betting on.

So, place the bet.

The Flywheel Spins

So what happens after winning the bet?

An investment turning into a trillion-dollar company is just the first layer of return.

It begins to rewrite a city's industrial structure. The capital market becomes more active, the supply chain becomes more complete, AI has real manufacturing scenarios, and young people have more reasons to stay.

The relationship between a city and its companies reverses at this point. In the early days, Suzhou helped enterprises get a seat at the table. Once the enterprises grow, they become Suzhou's reason to continue attracting new projects, talent, and capital.

Suzhou’s numbers look excellent. In 2025, total industrial output value above a designated size was 4.89 trillion RMB, ranking second nationally. The output value of high-tech industries grew by 6.7%, accounting for 56.2% of the total, an increase of 1.5 percentage points year-on-year. Twelve companies had their A-share IPOs, ranking first nationally, surpassing Beijing, Shanghai, Guangzhou, and Shenzhen. For every ten newly listed companies in the country, one comes from Suzhou. The per capita disposable income of all residents was 80,796 RMB, exceeding 90,000 in urban areas, trailing only Shanghai, Beijing, and Shenzhen nationwide – and this is just a prefecture-level city.

Among the top ten cities by GDP, Suzhou's service sector accounts for only 52.9%, the lowest. While other major cities have long been transitioning towards a service-based economy, Suzhou is still relentlessly building things.

Suzhou has 160,000 industrial enterprises, covering 34 major industrial categories and 514 sub-categories. It is one of the most complete industrial centers in the country and the world. For industrial AI, this means data – high-quality, continuous, scaled production data generated daily on the factory floor. Other cities trying to develop industrial large models might first need to search for data. But Suzhou doesn't need to search. Its factories are the data source themselves. Models trained here can return to the same production line for validation and iteration. The production line is both the source of AI learning and the endpoint where AI generates profit.

With a base of 4.89 trillion RMB, even if efficiency gains from AI are only one or two percentage points, each point represents hundreds of billions in absolute incremental value.

Suzhou has also turned computing power into a kind of public service. The industrial park established a public computing power service platform. Companies submit their requirements, get matched with providers within 24 hours, sign a contract within a week, and even receive a 20% subsidy. The platform's operations manager describes it like ordering takeout: choose your flavor, portion, and delivery time, then place a single order. For a small startup that can't afford the computing power of big tech, being able to "order takeout" can mean the difference between life and death.

Once the flywheel gets spinning, what does the average person gain?

Let's look at some figures. In 2025, the city added 415,000 new urban jobs, accounting for nearly 30% of Jiangsu Province's total. Furthermore, 283,000 college graduates chose to stay in Suzhou.