Every day when you unlock your phone, that small hole at the top of the screen projects tens of thousands of invisible light dots onto your face—they assemble a three-dimensional model of your face and confirm that "it's really you." Apple phones typically use chips from American companies, while among Android phones that support 3D facial recognition, most of the light comes from a company in Changzhou—Vertilite.
Now, this company is going public. But its story is far more complicated than "Android supplier": it has installed "eyes" on new energy vehicles, and now it wants to squeeze into the server rooms of AI data centers. Three migrations in ten years, cumulative shipments exceeding 900 million units, yet the revenue on its books still cannot support the valuation rumored by the outside world.
Three Migrations of a Chip: From Unlocking Phones, to Seeing the Road, to Connecting AI
In 2017, Apple removed the Home button from the iPhone X, at the cost of adding a small hole on the front.
After that, the unlocking action of billions of people worldwide changed from "press once" to "take a look." Inside that small hole, a laser chip called "VCSEL" projects tens of thousands of invisible infrared light dots onto your face; the sensor reads the deformation of the reflected light and reconstructs a micron-level three-dimensional face model.
So, you think it's the phone that recognizes you, but in fact it's tens of thousands of light dots that have measured your face.
Apple has Apple's light source. On the Android side, the light comes from Changzhou—Vertilite. The company's hardest label is "the light source foundation of the Android chain": according to industry media reports, Vertilite accounts for more than 70% of shipments in the Android camp's 3D facial recognition module field; earlier, in 2023, general manager Chen Xiaochi gave a figure in an interview with an industry media outlet of "a market share of over 80% in areas such as Android phone Face ID."
Founder Zhao Li was born in 1988 and went to Stanford around 2012 to pursue a doctorate in materials science, studying under James Harris, a member of the U.S. National Academy of Engineering. During his doctoral studies, he and his team produced the world's first long-wavelength VCSEL laser that emits light continuously at room temperature.
In those years, 3D facial recognition was becoming a new standard feature of smartphones, but domestic consumer-grade high-end VCSELs were almost nonexistent, and the global high-end market was monopolized by foreign giants Lumentum and Coherent. In 2015, Zhao Li and several other Stanford PhDs returned to China to start a business; the company was initially called "Shanghai Zonghui Optoelectronics," and in 2018 it moved its headquarters to Wujin, Changzhou.
In the first three years of the startup, they bet almost all their resources on underlying R&D, and the electro-optical conversion efficiency of their first 850-nanometer chip reached first-tier levels right away. In 2019, Vertilite mastered epitaxial growth and became the first company in China to mass-produce VCSELs, with shipments exceeding 10 million units that year.
Once the phone business was stable, they set their sights on automobiles.
Automotive-grade is a battlefield of another order of magnitude. Consumer-grade chips in phones only need to withstand temperatures up to 85°C, while automotive-grade chips must endure from minus 40°C to 125°C; consumer-grade has a lifespan of three to five years, while automotive-grade must last fifteen years; consumer-grade can make money with a 90% yield, while automotive-grade requires over 95%—because a single LiDAR contains hundreds of VCSELs, and if one fails, the entire LiDAR may become ineffective.
Vertilite spent two and a half years obtaining A
EC
-Q102 automotive-grade certification and IATF 16949 system certification. In some areas it has even begun to overtake: its anti-reflection VCSELs supply RoboSense, Hesai Technology, and Waymo, with annual installations of supporting LiDAR exceeding one million units, and a domestic market share of over 60% in light sources for new energy vehicle LiDAR.
But the automotive-grade market quickly became fiercely competitive too, and Zhao Li had to find another battlefield—this time, AI data centers. The reason is also simple: the more GPUs in an AI cluster, the less traditional copper wire can keep up with data transmission, and "optics in, copper out" has become an industry consensus. And to build 800G and 1.6T optical modules, the light source for short-distance optical interconnection still has to be VCSEL.
In August this year, Vertilite splurged 5 billion yuan to build a new high-end laser chip production line in Changzhou, focusing on indium phosphide.
And the latest news shows that the company has submitted coaching filing to the Jiangsu Securities Regulatory Bureau, with Guotai Haitong serving as the coaching institution.
900 Million Units and Hundreds of Millions in Revenue: A Company with Great Imagination but Still Small Revenue
If you only look at the first half, you and I would both think this is a hidden champion quietly making a fortune.
Especially after news spread that it is undergoing IPO coaching, "cumulative shipments exceeding 900 million units," "domestic consumer electronics VCSEL share exceeding 70%," and "valuation exceeding 15 billion yuan."
But if you flatten the timeline, a strange phenomenon appears: the company's growth rate does not match its revenue.
In 2023, general manager Chen Xiaochi said in an interview with an industry media outlet that the company's cumulative shipments were "over 90 million units"; by 2025–2026, the wording of industry associations and third-party institutions became "600 million units" and "700 million units." And the most popular recent claim is "over 900 million units."
The valuation is even more exaggerated. In Tianyancha's business database, Vertilite's "valuation exceeding 15 billion yuan" has been widely circulated and seems to have become a "market consensus." Yet the only figure with legal effect is hidden in a listed company announcement: last June, Dongyangguang increased its capital by 90 million yuan, corresponding to a pre-investment valuation of 3.285 billion yuan.
15 billion and 3.285 billion—a difference of more than four times.
At the revenue level, Vertilite's scale is not large. In 2024, Vertilite's revenue was 125 million yuan, with a net loss of 58.2982 million yuan. In the first quarter of 2025, revenue was 41.2343 million yuan, with a net loss of 15.3154 million yuan.
900 million units and hundreds of millions in revenue. This is the story tension of this company: it started with "cheap volume" and accumulated a number that sounds shocking, but this number has never been converted into money.
Looking at the shareholder register is even more intriguing: Hubble Investment, Xiaomi Yangtze River Industry Fund, BYD, DJI, Ant Group, and even competitors Hesai Technology and RoboSense. Phone makers, automakers, LiDAR makers, drone giants, payment giants—Vertilite has gathered almost the entire industry chain.
But the matter of "shareholders as customers" is something the company must seriously address at the time of IPO, and regulators will certainly ask: investment does not equal purchase orders, and the actual sales revenue from these shareholders has so far not been fully disclosed.
In addition, this industry has a crueler side: while shipments surge, unit prices continue to decline.
For example, peer Changguang Huaxin's revenue from VCSEL and optical communication chips in 2025 was 41.1938 million yuan, a year-on-year increase of 1036.79%—which sounds impressive, but its gross margin decreased by 37.5 percentage points year-on-year to 12.81%.
Another company, Dugen Core, is the same: in the first half of 2026, revenue from VCSEL chips and modules was 12.1191 million yuan, accounting for only 3.13% of its main business revenue.
In other words: across the entire domestic VCSEL optical communication track, publicly disclosed revenue is still at the tens-of-millions level.
So Vertilite must change its story. The 5 billion yuan investment in August this year is the new story.
Zhao Li himself has also explained: traditional gallium arsenide processes are mature and can still rely on foundries; but in the frontier indium phosphide area, "a standard and mature foundry division-of-labor ecosystem has not yet formed globally," and materials, design, and processes are bound together, "highly dependent on repeated debugging and accumulation of frontline engineering experience." If it continues to rely purely on foundries, "the development cycle will be greatly lengthened, and the core process know-how will be difficult to truly master."
Simply put, Vertilite is no longer an asset-light design company; it must build its own production line.
But will its competitors give it time? At present, Coherent is betting on the world's only 6-inch indium phosphide production line, planning to double its internal InP capacity by the end of 2026 and land it a quarter earlier; Lumentum's CEO said publicly in April this year that because it is "increasingly unable to keep up with demand," all capacity through the end of 2028 has been booked out. In the high-end 200G EML field, orders for overseas leaders have been scheduled beyond 2027, while domestic manufacturers as a whole are still in the verification and small-batch stage.
So the 5 billion yuan buys not just a production line, but an admission ticket. Whether it can secure a stable seat still depends on yield.
This article comes from the WeChat public account "Phoenix Technology," author: Phoenix Technology
