Stratistics sees AI data center silicon photonics at $14.6 billion by 2034

The forecast rests on a straightforward pressure point: AI models, especially large language models and high-performance computing, push far more data between GPUs, processors, memory and network gear than electrical links handle efficiently. Stratistics argues that optical connections ease that bottleneck, and it cites the International Energy Agency's finding that data centers consumed about 415TWh of electricity in 2024, roughly 1.5% of global power use, as context for why power-efficient transmission matters.

The report's segmentation points to where growth is expected to concentrate. Optical transceivers, which convert electrical signals to light and back, are projected to hold the largest share, while AI inference infrastructure is expected to grow fastest as generative AI services and real-time applications spread. Higher-speed links such as 800G and 1.6T are treated as increasingly important, alongside newer designs like co-packaged optics.

On the supply side, the report names Intel, Cisco, Broadcom, NVIDIA, Marvell and others as competitors, and notes that manufacturing cost and complexity, plus competition from existing electrical interconnects, could slow adoption. It also points to North America's concentration of cloud and chip companies and to Asia-Pacific's investment in AI and data centers as the geographic split behind the numbers.