NextFin News - Ligent Technologies Inc. is set to begin trading on the Hong Kong Stock Exchange on Sept. 22 after pricing a $727 million initial public offering at HK$32.96 a share, a debut that puts the AI data-center optics supplier at a roughly $4.1 billion valuation and tests whether investors will keep paying up for the infrastructure behind the artificial-intelligence buildout.
The listing arrives two months after Zhongji Innolight, a larger rival in the same optical-transceiver market, raised Hong Kong's biggest IPO in seven years only to close its first trading day lower. Ligent's debut is the latest stress test for a sector where the long-term demand story is strong but the short-term appetite for new supply is proving fickle.
The Deal: $727 Million for the Pipes Inside AI Data Centers
According to the company's filing with the Hong Kong stock exchange, Ligent is offering 172 million new shares at HK$32.96 apiece, raising about HK$5.7 billion ($727 million). All of the shares are primary shares, so the proceeds go to the company rather than to existing shareholders, and the underwriters hold a 15% over-allotment option that could add up to 25.8 million more shares. At the offer price, the company carries an implied market capitalization of about HK$32.4 billion ($4.13 billion).
Nearly 30 cornerstone investors have committed to buy 47% of the base offering, equivalent to roughly $340 million. The list includes Primavera Investment Fund, GigaDevice, Amlogic Hong Kong, Mirae Asset Securities, PAG, the ORIX-backed Turquoise Hime, Barings, GF Fund and E Fund. Heavy cornerstone participation is a double-edged signal: it reduces the free float that can hit the market on day one, but it also means almost half the deal was placed with institutions that are locked in rather than testing open-market demand.
Headquartered in San Jose, California, and controlled by China's Hisense Group, Ligent makes optical transceivers, chips and network terminals — the hardware that converts electrical signals into light and back again so data can move between servers and racks. About 70% of its revenue comes from datacom transceivers, the modules that plug into switches inside AI data centers and route traffic between GPU clusters. In the first half of 2026, revenue rose 27.7% to 5.39 billion yuan ($803 million) from 4.22 billion yuan a year earlier, and profit climbed nearly 30% to 661 million yuan, driven by higher datacom-transceiver sales.
The growth is real. The question is whether it is already priced in.
The Structural Case: AI Is Forcing Copper Out of the Data Center
The bull case for Ligent does not rest on sentiment. It rests on physics. As AI models scale from thousands of chips to tens of thousands, the volume of data moving between processors has outpaced what copper wiring can carry over useful distances. Optical interconnects — the category Ligent sells into — are no longer a premium option; they are becoming the default architecture for AI clusters.
Yole Group, a market-research firm that tracks the sector, has described the moment as the biggest structural shift in optical interconnects in 40 years, driven by AI clusters, silicon photonics and new form factors. The firm forecasts the optical-transceiver market will reach $112 billion by 2031, up from about $23 billion in 2025, with AI data centers driving average annual growth of 30%. Lakshman Srinivasan, a technology and market analyst for photonics and lighting at Yole Group, put the shift in plain terms:
The optical transceiver market is entering a new growth cycle. As AI clusters become larger and more distributed, optical connectivity is evolving from a networking component into a strategic infrastructure technology, creating significant opportunities across the value chain.
The numbers behind that assessment are striking. China's AI optical-transceiver market grew from 1.6 billion yuan in 2021 to 16.7 billion yuan in 2025, a compound annual growth rate of 79.6%, according to the Frost & Sullivan industry report in Ligent's prospectus. The same report projects the market to reach 99.0 billion yuan by 2030, implying 42.8% annual growth for the rest of the decade.
Global AI infrastructure investment tells the same story: it rose from 218.0 billion yuan in 2021 to 2.39 trillion yuan in 2025, an 82.0% annual pace, and China's AI computing capacity is forecast to expand from 741.3 EFLOPs in 2025 to 3,633.4 EFLOPs in 2030. Every increment of that capacity needs more optical links. This is not a cyclical inventory restock; it is a change in the material composition of the data center.
That distinction matters for how investors should read Ligent. A cyclical upswing reverts when demand normalizes. A structural shift re-prices the entire addressable market and leaves smaller suppliers with a longer runway. Ligent's 2025 AI optical-transceiver revenue of 1.1 billion yuan, against a global AI optical-transceiver market of 71.8 billion yuan, gives it a 1.6% global share — eighth among specialized suppliers worldwide and seventh in China, by the company's own commissioned industry figures. For a buyer of the structural thesis, a 1.6% share is not a weakness; it is the growth option.
The Cyclical Case: The IPO Window Just Punished the Category Leader
The bear case is simpler and more immediate: the market has already eaten this lunch. In July, Zhongji Innolight raised HK$53.4 billion ($6.8 billion) in what was Hong Kong's largest IPO in seven years. Shares priced at HK$980 and closed the debut session at HK$960, down 2%, after falling as much as 10% intraday. That was not a liquidity hiccup; it was a signal that even the strongest name in the sector could not escape a broader cooling in AI-infrastructure enthusiasm. On the same day, Innolight's Shenzhen-listed shares slipped 9.15%, underscoring that the pressure was market-wide rather than specific to the Hong Kong listing.
There is a second, starker data point. Lightelligence, a Shanghai-based optical-computing company, priced its Hong Kong IPO at HK$183.20 in April and surged as much as 408% intraday to HK$930 before closing up 383.6% at HK$886, raising about HK$2.4 billion. Two AI-optics listings, two opposite outcomes. The dispersion is the story: the market is no longer paying for the category; it is underwriting the specific company.
That selectivity is uncomfortable for Ligent at a $4.1 billion valuation. Innolight, the category leader, commands a market value roughly 30 times larger and still fell on debut. Ligent's revenue base — 5.39 billion yuan in the first half — is an order of magnitude smaller, and its customer concentration and geographic mix are less diversified. The company generated most of its recent growth from datacom modules tied to AI capital spending, which means its fortunes move in lockstep with a spending cycle that has already delivered a decade's worth of growth in about four years.
The cornerstone structure also cuts against the debut. With 47% of the offering locked up, the tradable float is thin. Thin floats can amplify gains when demand is strong, but they also mean a small amount of selling pressure can produce an outsized move. In a market that just handed Innolight a 2% first-day loss, "thin" is more often a risk descriptor than a catalyst.
What the Market Is Pricing — and What It Is Missing
The conventional read of this listing is straightforward: AI data-center capital spending stays strong, optical-transceiver demand follows, Ligent grows into its valuation. That view is probably right about the direction and probably wrong about the risk. It treats the sector as a single bet when it is actually two bets layered on top of each other.
The first bet is structural and has a long half-life: the migration from copper to optics inside AI clusters is durable because it is driven by power and bandwidth constraints, not by procurement fashion. The second bet is cyclical and short-dated: that the IPO window for Chinese technology listings stays open, that AI capital-spending guidance from hyperscalers does not disappoint, and that Hong Kong retail demand for new issues does not evaporate between pricing and debut.
Investors are being asked to pay a structural multiple for a company whose near-term stock performance will be decided by the cyclical layer. That mismatch is where the risk sits. If hyperscaler capital spending holds and the debut trades flat or better, the structural narrative reasserts itself and the valuation looks cheap in hindsight. If spending guidance wobbles or the debut stumbles, the 1.6% market share that bulls cite as an option becomes, in the short term, a reason the stock is untestable — there is not enough scale yet to prove the company can convert industry growth into company growth.
The Counter-Thesis: Why the Skeptics May Be Reading the Wrong Signal
The strongest argument against the cautious view is that the Innolight debut was an anomaly, not a trend. Innolight listed into a global AI sell-off that dragged its Shenzhen-listed shares down 9.15% on the same day; the Hong Kong debut loss was a beta event, not a verdict on optical transceivers. By that logic, comparing Ligent's debut to Innolight's is comparing two different market regimes.
There is weight to that objection. The Lightelligence surge shows that Hong Kong investors are still capable of enthusiastic repricing when a story is compelling and the float is tight. Ligent's 47% cornerstone lock-up, which looks like a risk in a weak tape, becomes an advantage if sentiment turns: with less than half the shares immediately tradable, a modest bid can move the price disproportionately higher.
But the counter-thesis has a hole. Innolight's revenue grew 60.3% in 2025 to 38.24 billion yuan, and its net profit more than doubled to 11.58 billion yuan, with 57.3% of revenue coming from the United States in 2025 and 61.7% in the first quarter of 2026 — a far more robust fundamental profile than Ligent's. If the market was willing to mark down the stronger company on day one, it is not clear why it would mark up the smaller, less diversified one on the same structural theme. The burden of proof has shifted to Ligent's debut tape.
What to Watch: The Debut and the Signal That Would Break the Thesis
The near-term read is simple and observable. If Ligent's shares hold above the HK$32.96 offer price through the first week of trading, it would indicate that cornerstone demand was genuine and that the market is still underwriting AI-infrastructure listings at premium valuations. A close below the offer price, especially on heavy volume, would confirm that the Innolight debut was a pattern rather than an exception.
Beyond the debut, three catalysts matter. First, hyperscaler capital-expenditure guidance for 2027: any downward revision to AI data-center spending would hit datacom-transceiver suppliers first and hardest. Second, the pace of the transition to 800G and 1.6T modules: Ligent's ability to move up the value chain determines whether it can defend margins as the market shifts. Third, customer concentration: the prospectus flags reliance on a narrow set of large buyers, and any loss or delay from a top customer would weigh on a company whose revenue is 70% tied to one product category.
The falsifying signal for the structural thesis is specific: if China's AI optical-transceiver market fails to grow at least 30% in 2026 — well below the 42.8% annual pace projected through 2030 — the "structural shift" narrative loses its supporting data and the sector rerates as a cyclical capital-spending play. For the debut-specific call, the signal is equally concrete: a first-week close below HK$32.96 on volume exceeding the 172-million-share offering would indicate that the IPO window for AI-infrastructure names has narrowed faster than the fundamentals.
Bottom Line
Ligent is a legitimate beneficiary of a genuine structural shift, but legitimacy is not the same as value at the offer price. The AI data-center buildout is a multi-year tailwind; the IPO reception is a one-day verdict. Investors buying the debut are not just buying optical transceivers — they are betting that the market will keep distinguishing between the sector's leaders and its challengers, and that Ligent has done enough to be counted among the former.
Three scenarios frame the path from here. In the base case, the shares trade roughly flat around the HK$32.96 offer price in the first week, reflecting genuine cornerstone demand offset by a cautious broader tape for AI-infrastructure listings. In the upside case, a close above HK$38 — roughly 15% over the offer price — on volume exceeding the 172-million-share offering would signal that the Lightelligence-style appetite for tight-float AI optics names has returned. In the downside case, a first-week close below HK$30 would confirm that the Innolight debut was a pattern, not an exception, and that the market is discounting secondary names in the supply chain ahead of the leaders.
The debut tape will answer that quickly. Until then, the prudent read is that the industry is structural, the window is cyclical, and at $4.1 billion Ligent is being asked to prove the former before the latter closes.
Explore more exclusive insights at nextfin.ai.
