Olix will ship its first photonic semiconductor product in 20271, joining a global race where Nvidia projects $1 trillion in cumulative chip sales through the same year2. Photonic chips transmit data via light rather than electricity, targeting energy efficiency gains for AI inference operations that now span data centers worldwide.
The 2027 timeline mirrors production schedules from specialized chip developers across North America, Europe, and Asia. Language Processing Unit makers focused on transformer models1 share similar launch windows, while European semiconductor initiatives and Asian foundry partners expand capacity for non-traditional architectures.
Infrastructure bottlenecks drive architectural diversity. Training frontier AI models requires tens of thousands of interconnected accelerators, creating power and cooling challenges from California to Singapore data center hubs. Inference workloads—running deployed models at scale—present different constraints around latency and per-query energy consumption that vary by regional electricity costs and grid capacity.
Established players scale conventional production globally. Micron is acquiring High-Bandwidth Memory fabrication facilities2 to supply GPU components, while Meta's $12 billion AI infrastructure commitment spans international partnerships. Amazon's Tranium chips and Google's TPUs represent hyperscalers building custom silicon to reduce dependence on merchant suppliers dominated by U.S. and Taiwan-based manufacturers.
Photonic interconnects address data movement limits in multi-chip systems where inter-processor bandwidth constrains scaling regardless of geographic location. This positions Olix's technology for deployment scenarios from edge inference in bandwidth-constrained markets to hyperscale data centers in fiber-rich regions.
The trillion-dollar forecast encompasses accelerators plus supporting infrastructure: memory, networking silicon, and packaging technologies enabling multi-die integration. This broader value chain explains simultaneous capacity expansion across continents—from memory makers in South Korea to interconnect specialists in Israel to packaging facilities in Malaysia.
Sources:
1 Source, "While OpenAI Shattered Records, Robotics and Semiconductor Startups Quietly Added The Most New Unicorns In February"
2 Olix, via analysis
