AI's need for compute density forces data centers to shrink data movement from kilometers to millimeters, triggering a full-stack redesign across power, cooling, networking, and rack archit…

As AI compute density increases, electrical interconnects hit physics limits; optical I/O and co-packaged optics become mandatory to move data at millimeter scale with acceptable power and…

Leading AI labs are tightly co-designing model architectures with custom silicon (XPU/ASICs), creating a feedback loop where model specs inform chip design and chip capabilities inform mode…

New power architectures eliminating voltage step-down losses, combined with behind-the-meter generation (SMRs, fusion), are necessary to support gigawatt-scale AI clusters and represent a p…

Model architectures are shifting from compute-heavy to memory-heavy (mirroring human brain), creating structural memory shortage as 3-4 year fab lead times clash with exponential demand gro…

Error-corrected million-qubit quantum computers are on a 2030 timeline, requiring decade-early capital allocation; Kim has been invested since 2019, treating quantum as asymmetric frontier…

Multiple frontier technologies (quantum, SMRs, fusion, 800V power, orbital compute) share a ~2030 commercialization target, suggesting a step-function infrastructure shift where space-based…

Small modular reactors and fusion are on parallel 2030 timelines to provide behind-the-meter, carbon-free baseload power for gigawatt-scale AI clusters, de-risking grid dependence.

Decades of VC neglect caused chip company consolidation into few survivors per category, creating non-commodity pricing power with highest sector margins; AI demand now exposes this structu…

China's 130+ robotics companies and 30-40 near-term IPOs reflect manufacturing scale advantages; the winning paradigm may mix Chinese hardware bodies with Western foundation model brains, a…

Autonomous agents generate massive incremental data volumes requiring new memory/storage architectures and database paradigms, creating downstream tailwinds for data infrastructure and memo…

Value accrues to companies in the 'token flow' — either creating tokens (compute/foundation models), serving tokens (inference clouds), or repackaging tokens with proprietary context (verti…

Enterprises must embody cumulative knowledge into a context/ontology layer (à la Palantir) to enable agent orchestration over proprietary data; this becomes the new defensible moat above co…