Demand for high-speed optical networking is rising alongside the rapid expansion of generative artificial intelligence, with industry players anticipating significant growth through 2027. In China, the emergence of large-scale “superpods” — advanced AI server clusters — is expected to fuel adoption of 800G optical modules from 2026, while global markets, led by US cloud providers, are forecast to see strong double-digit growth.
However, supply conditions remain uneven. Some manufacturers report shortages of certain laser components, particularly electro-absorption modulated lasers, while others highlight stable supply due to long-term procurement agreements. Meanwhile, new technologies are advancing, with innovations in module formats and materials aiming to support faster data transmission and lower costs.
China’s AI infrastructure push is also reshaping server demand. Superpods are scaling rapidly, with configurations expanding up to 1,024 GPUs. Unlike the US, China’s focus is more heavily weighted toward inference rather than training, reflecting its reliance on open-source foundation models. Government backing, including targets to achieve 90% generative AI adoption by 2030, is further accelerating investment.
Despite strong demand, supply challenges persist. Companies report tighter availability of CPUs compared to memory, though domestic alternatives are emerging across chips, memory, and manufacturing.
At the same time, local firms are developing new AI chip architectures tailored for inference, aiming to improve energy efficiency and flexibility. Industry participants remain optimistic, citing both market expansion and policy support.
Beyond hardware, AI applications are expanding rapidly. Businesses are deploying AI agents to enhance manufacturing efficiency and streamline sales, while consumer demand is shifting toward multi-modal tools and personalised assistants.
Looking ahead, emerging technologies such as quantum computing and autonomous transport are also gaining traction, with robotaxi fleets expected to expand significantly in the coming years as costs fall and capabilities improve.
