Photonicsense Patent Portfolio
Four patents addressing the three fundamental bottlenecks in AI infrastructure — memory bandwidth, compute efficiency, and thermal limits — mapped against the same demand curves driving our investment thesis
The 30-Second Version
What this portfolio does, in plain English
AI hardware has three bottlenecks: memory is too slow (GPUs waste time managing data instead of computing), compute burns too much energy (electrons through silicon is inherently lossy), and everything overheats (3D-stacked chips and dense racks hit thermal walls). These four patents solve all three bottlenecks — and they work together as a platform, not just individually. The same companies buying HBM from SK Hynix and optical transceivers from Lumentum are the buyers for these patents.
Who Would Buy These — And Why They'd Pay
Each buyer has a specific problem these patents solve
How the Patents Connect to Our Investment Holdings
The same demand curves that drive SK Hynix, Lumentum, and Coherent also drive patent value
Portfolio Valuation
Conservative licensing model through aggressive acquisition scenario
Commercialisation Timeline
Priority actions for maximum value capture
Key Risk: OpenCxMS Technologies filed 15 provisional patents in February 2026 in autonomous safety territory overlapping the IOC. PCT international filing is urgent to establish priority. The SMC, VPPNN, and Dean Vortex patents have no known competitive filings.
Portfolio Combined Valuation: $2.2B base / $5B bull by 2030 across all licensing streams.
The patents become more valuable as the AI infrastructure buildout continues — the same macro thesis that drives the investment portfolio drives patent value. This is a dual exposure strategy: equities capture market-level returns while IP captures technology-level rents.