The upcoming IPOs of Montage Technology and Axera Semiconductor, both scheduled for Hong Kong, provide valuable insight into Chinese strategy. More than financial operations, these IPOs show a coherent industrial policy, where artificial intelligence is not understood as a complete productive chain, going from silicon to the machine in situ.
Through these two issues, the markets do not finance a fashionable sector. They accompany an assumed industrial architecture, based on the continuity of technological links and on the acceptance of long temporalities, sometimes incompatible with the classic standards of Western markets.
Two trajectories, the same logic of continuity
Montage is an installed actor. Founded in 2004, the company designs chips for interconnection and optimization of data flows at the heart of data centers and AI accelerators. It displays significant profitability and a high valuation, driven by the rise in computing infrastructure. Its listing in Hong Kong, accompanied by a significant discount compared to its Shanghai price, reflects a desire to become more open to international investors and diversify its shareholder base.
Axera, conversely, embodies a more upstream phase. Created in 2019, the company specializes in embedded AI inference chips, intended for cameras, industrial equipment or intelligent vehicles. Its activity is progressing, but its losses are increasing, reflecting a sector still in structuring, highly capital intensive and technologically demanding.
From computing infrastructure to embedded intelligence
Taken together, these two players cover a broad technological spectrum. Montage operates at the heart of centralized infrastructures, where data flows, accelerators and massive computing capacities are concentrated. Axera is positioned at the other end, closer to uses, where AI must operate in real time, under energy constraints and without permanent dependence on the cloud.
After a phase largely dominated by the development of centralized models and hyperscale infrastructures, the challenges are now embedded inference, edge computing and distributed intelligence. Intelligent cameras, industrial robots, vehicles, surveillance or control systems, AI is gradually leaving data centers to become part of concrete objects, inserted in physical environments.
Financing these players therefore amounts to financing industrial capacities to ensure future production.
Hong Kong, financial platform for an industrial strategy
In a context of increased restrictions on American markets and persistent technological tensions with Washington, the choice of Hong Kong as a listing location is not neutral. It makes it possible to mobilize international capital while remaining integrated into the Chinese ecosystem.
The volume of IPOs linked to AI and semiconductors, concentrated at a particularly active start to the year, suggests that Hong Kong is now being used as a financing lever for technological upscaling.
A systemic approach rather than an isolated bet
These IPOs do not paint the portrait of two isolated companies, but that of a system and demonstrate the support that China has for the entire AI chain, from semiconductor to machine.
This approach contrasts with a more fragmented vision, often observed in Europe or the United States, where capital is primarily focused on software and models, leaving hardware to depend on a few dominant suppliers. It assumes acceptance of industrial risk, long cycles and high capital requirements.
From this perspective, China’s AI strategy appears unambiguous, to go beyond the most efficient model, and address the entire chain of silicon transformation to uses.
