CXMT and Unitree Queue for IPO: Brokerages Embrace a Boom Year for Hard-Tech Investment Banking
Miles Bennett
CXMT, Unitree and YMTC are pushing toward IPOs, and the STAR Market's mandatory co-investment rule is turning brokerages from fee-collecting middlemen into holders of hard-tech equity — buying a brokerage stock is becoming a discounted bet on a tech-stock basket.
Why are brokerages suddenly trading like tech stocks?
The STAR Market has a mandatory rule: the sponsoring brokerage must co-invest its own capital in every IPO it underwrites, at 2 % to 5 % of the offering, locked up for 24 months. This means → brokerages no longer just collect a one-time underwriting fee — they ride the stock's upside and downside.
In plain terms = they used to be real-estate agents earning a commission; now the rules force them to buy the house too — if the price rises, they profit alongside.
Of the 30 STAR Market IPOs since early 2025, over 70 % have generated co-investment paper gains exceeding total underwriting fees. The brokerage sector trades at just 1.20× book value, yet average daily A-share turnover has hit ¥3.19 trillion — the gap between strong fundamentals and depressed valuations is drawing sustained inflows.
How much can co-investment actually earn?
Guotai Junan estimates 2025 STAR Market IPOs averaged 200 %–260 % gains on debut. If 2026 STAR Market IPO volume reaches ¥60 bn at an average 3 % co-investment ratio, the industry commits roughly ¥1.8 bn — yielding an estimated ¥4.5 bn in paper gains.
Peak quarterly co-investment profit for a top brokerage could exceed ¥7 bn. This means → co-investment gains alone can swing a major brokerage's quarterly earnings.
Sun Ting, chief non-bank analyst at Soochow Securities, notes the investment-banking model has shifted from "collect a fee" to a full life-cycle of "industry insight → deal sourcing → capital empowerment → value realisation." In plain terms = the deeper a brokerage understands an industry and the earlier it invests, the more it earns — the era of pure licence-based fees is ending.
The CXMT IPO — how thick is the profit chain?
CXMT (长鑫科技) is China's No. 1 and the world's No. 4 DRAM maker — DRAM is the memory chip inside every phone and PC. It plans to raise ¥29.5 bn, the second-largest STAR Market offering ever after SMIC. Q1 2026 net profit hit ¥24.76 bn, up 1,688 % year-on-year.
Kaiyuan Securities maps three revenue layers: underwriting — at a 1.5 % fee, CICC and CSC each earn roughly ¥220 mn; co-investment — under a ¥2 tn market-cap scenario, each sponsor's paper gain is about ¥5.3 bn, rising to ¥8.3 bn at ¥3 tn; direct investment carries the most upside — CMS and Huaan Securities hold the largest stakes on a look-through basis.
Combining direct and co-investment at a ¥2 tn market-cap assumption: CMS's theoretical gain is roughly ¥13.2 bn, CSC about ¥7.6 bn, Huaan about ¥6.9 bn, CICC about ¥5.4 bn. This reflects how a single mega-IPO can lever brokerage profits far beyond anything traditional underwriting ever delivered.
Where are we in the investment-banking cycle?
Full-market IPO volume in 2025 was ¥131.8 bn — just 22 % of the 2022 peak of ¥586.9 bn. The 2026 forecast is roughly ¥200 bn, still in the low band after the last boom. This means → the recovery is real, but we are in the early half of a climb from the bottom, not near a new peak.
Concentration is accelerating fast: the top five brokerages' IPO market share rose from 52 %–55 % in 2020–2022 to 73 % in 2025. Their IB profit margin recovered from 19 % in 2024 to 37 % in 2025 — still below the roughly 50 % highs of 2020–2021.
CICC's "big IB" segment now contributes about 19 % of firm-wide profit, up 3 percentage points year-on-year. CSC, Guotai Junan, CITIC Securities and Huatai each contribute about 15 %, boosted mainly by direct- and co-investment units swinging from loss to profit.
What else is in the pipeline?
Unitree Robotics plans to raise ¥4.2 bn and has filed with the CSRC, sponsored by CITIC Securities. YMTC completed its guidance filing on 19 May 2026, co-sponsored by CITIC Securities and CSC.
xFusion Digital Technologies plans to raise ¥8 bn and is in the inquiry stage. Enflame Technology plans to raise ¥6 bn and has filed with the CSRC. This means → these four projects plus CXMT alone total nearly ¥48 bn in planned fundraising — an historically rare density of hard-tech IPOs.
On deal count, CITIC Securities and Guotai Junan each had 23 non-BSE IPO mandates from 2025 to May 2026, tied for the top. Cumulative co-investment gains (June 2024–June 2026): CITIC Securities ¥3.17 bn, Guotai Junan ¥1.88 bn, CSC ¥1.33 bn.
How far can this thesis run?
Kaiyuan Securities sees three layers of support: short-term earnings elasticity from co-investment gains, mid-term project reserves as CXMT, YMTC and others list, and long-term structural uplift in IB return on equity.
In plain terms = there is money to earn now, a queue of deals coming, and the business model itself is upgrading — stacking these three layers is why the market is willing to re-price brokerages.
One caveat: co-investment gains are locked for 24 months and will not flow to the income statement all at once — actual realisation depends on the share price at unlock. The pace at which hard-tech IPOs actually land remains the key test of whether this thesis holds.
Content is for reference only, not financial advice.