Moore Threads Technology Co., Ltd. (NASDAQ: N/A) experienced a 20% share price drop on September 7th, followed by further losses exceeding 9% in intraday trading on September 8th before trimming those declines. The catalyst was the unlocking of 25.77 million restricted shares from its initial public offering (IPO) network placement on September 7th. The company's total floating share capital has since expanded by 85%, according to its official statement.
The company has responded to inquiries by confirming its operational performance continues to improve, with products used for AI large model training and inference already sold to internet companies, major AI model developers, AI infrastructure providers, and telecom operators. The precise operational and financial figures will be disclosed in its official announcements. Prior to this unlock event, Moore Threads Technology Co., Ltd. only had 30.23 million shares in public float; this figure has now risen to approximately 56 million shares.
Based on the September 4th closing price, the newly unlocked shares were valued at RMB 13.39 billion. After this batch of shares is released, the public float will represent 11.9% of the company's total share capital. A more significant test looms on December 7th when 186 million shares held by early IPO investors (representing 39.55% of total share capital) are scheduled for unlock.
The sector faces similar pressures. Metax Integrated Circuits (Shanghai) Co., Ltd. (NASDAQ: N/A) is set to release 13.97 million IPO institutional placement shares on September 17th, expanding its public float by 75%. This will be followed by another 10.92 million shares from original shareholders due on December 17th. Metax Integrated Circuits saw its share price decline over 9% on September 7th and continued to fall more than 1% in the subsequent session.
Beyond the immediate market pressure from expanded share supply, the long-term trajectory for these GPU makers depends on their fundamental performance. In the first half of this year, the four major listed domestic GPU companies all reported substantial revenue growth, though their profitability painted a mixed picture. Moore Threads Technology Co., Ltd. posted the highest revenue at RMB 1.736 billion, marking a 147.42% year-on-year increase, while its loss narrowed significantly to RMB 11.56 million from RMB 271 million in the year-ago period.
BIREN TECH (HKEX: 06082) recorded first-half revenue of RMB 1.236 billion, up a remarkable 1,997.6% from last year. Its loss narrowed dramatically to RMB 377 million from RMB 1.6 billion a year earlier. Meanwhile, Metax Integrated Circuits (Shanghai) Co., Ltd. and ILUVATAR COREX both managed to swing to profitability in the period. Metax Integrated Circuits achieved revenue of RMB 1.324 billion, an increase of 44.67%, with net profit reaching RMB 612 million. ILUVATAR COREX posted revenue of RMB 946 million, up 191.6%, and net profit of RMB 106 million.
However, the profit improvements at Metax Integrated Circuits were largely driven by non-core business activities. In the first half, its net operating cash flow was negative RMB 1.297 billion, while its fair value change gains amounted to RMB 887 million, accounting for 105.75% of its total profit. ILUVATAR COREX similarly recognized RMB 760 million in fair value gains during the period.
Regarding market demand, a technology research head at UBS Securities China noted that demand for domestic AI chips continues to surge this year, which is reflected in the significant capital expenditure increases by domestic cloud providers. From the second half of this year through next year, computing power shipments domestically, whether in super-node configurations or other forms, are expected to maintain rapid growth trends.
Industry sources indicate that accelerator chips now account for more than one-third of AI infrastructure construction costs. Currently, domestic AI firms primarily deploy domestic computing power for inference tasks. For training purposes, domestic chips are mainly used for testing as a measure to safeguard supply chain security.
On the supply side, the same UBS analyst pointed to supply chain bottlenecks that could affect shipment schedules. While domestic advanced process manufacturing capacity is increasing, it does not yet equate to fully efficient production. Achieving good yields to meet domestic accelerator card demand will still require time. Beyond wafer fabrication as a potential constraint, HBM memory, capacitors, and PCB boards could also emerge as bottlenecks. Despite growing demand, many supply-side challenges remain to be resolved.
Comments