A definitive briefing on the financial, physical, and cognitive bottlenecks defining the AGI transition.
As we enter 2026, the AI ecosystem has undergone a fundamental phase transition. The era of "research curiosities" has been replaced by an industrial siege. Frontier labs are no longer just optimizing transformers; they are resurrecting nuclear reactors and building gigawatt-scale data fortresses.
This newsletter dissects the three "Great Filters" currently throttling progress: the Capital Gap, the Energy Wall, and the Labor Realignment. Our model suggests AGI remains targeted for 2028, provided these physical constraints are met.
Adjust global scaling vectors to calculate the probability of the "Hard Takeoff" scenario.
We are witnessing a $600 Billion front-loading of infrastructure. This represents a "leap of faith" in scaling laws. Currently, actual AI service revenue covers only 15% of annual CapEx.
Labs are spending $1B per training run in 2026.
Valuations rely on a "Margin Pivot" scheduled for late 2026.
AI compute demand is decoupling from standard civil infrastructure. 2026 is the year of "Grid Defection."
Direct on-site baseload generation for Level 4 labs.
Bypassing the queue via dedicated natural gas hookups.
Cluster density is now limited by liquid cooling physics.
The hardware battle between NVIDIA and Hyperscaler ASICs.
Replacement vs. Creation: The path to 2030.