With AI demand outpacing supply & domestic grid growth slowing, the US will soon consume more tokens than it produces. A basic model puts the switch to a net importer between 2028 & 2030.
The cost of an autonomous corporate break-in has fallen 99% in five months to $90. As attackers run expected-value calculations on AI compute, CISOs must architect for speed, scale, & breadth to drive the attacker's expected return below zero.
The ceiling for AI in 2027 : about 80gw. That number doesn't mean much to most people, but it's the upper bound on how fast AI can grow & how much revenue it can generate.
Anthropic's S-1/A reveals $17.0b of Q3 2026 revenue, $5.0b of short-term burst compute at $7 to $13 per GPU-hour, & $895.4b of total compute commitments.
AI has four layers : the application, the harness & context layer, the model, & the data center. Each faces a critical question about its future, & a shift at any one layer reshapes the other three.
Two CEOs now project $1t in revenue. Cloud growth is accelerating at $30b to $50b a quarter, but AI is eroding the legacy search & ad businesses that pay for the capex. The first $10t company will be the one that crosses that chasm first.
Arvind KC, Chief People Officer at OpenAI, on why efficiency is the wrong framing for AI, how companies democratize intelligence by amplifying their most valuable capability, & why EQ becomes the differentiator.
Security looks like the third AI market after coding & customer support. Attacks on banks, certificate authorities & open source libraries have never been greater, & every category of security work now carries heavy token consumption. Greg Brockman thinks security spend should vastly outstrip software spend.
Major build-outs run in two phases, per Carlota Perez: the infrastructure, and then the applications that pay for it. The internet took seven years to close the gap, and AI is closing it in about half the time. Whether that is circular depends on how large the circle is.
Granola's VP of Sales on why AI-era teams run smaller & quotas run bigger : eight-figure deals in two weeks, quarterly comp plans, & a sales pod that splits by layer.
