China's up to 100x cost advantage is reshaping the AI race
THE SO WHAT
A claimed up to 100x cost advantage on some Chinese AI workloads would turn "who has the best model" into "who can afford to run it at scale." If you’re building AI products, assume pricing pressure from low-cost regions and design for unit economics that can survive a race to marginal-cost inference.
READ THE SOURCE
MORE FROM THE WIRE
Applied AISources: Nvidia signs leases worth up to $50B for Hut 8's planned 1GW data center in Texas, hosting hundreds of thousands of chips; Hut 8 didn't name its client
Leasing up to $50B of capacity in a 1 GW Texas data center is Nvidia using its balance sheet to secure downstream demand and power, not just sell chips. For AI builders, compute access is consolidating into vertically coordinated ecosystems—your vendor choice is increasingly a bet on their capex strategy.
Jensen Huang says that AI job destruction doomers are 'exactly backward'
Framing AI as automating "tasks" not "jobs"—with examples in radiology and law—is an invitation to redesign roles around supervision and orchestration. Leaders should be mapping workflows at the task level now, because that’s where both automation upside and labor pushback will land.
Applied AIAI Wipes Out Customer Service Jobs at Microsoft, Uber, CBA
Call centers at Microsoft, Uber, and CBA seeing AI-driven job losses is the concrete version of "task automation"—frontline service work is the first labor shock. If you run large support operations, you need a plan for redeploying or exiting headcount over the next 12–24 months, not a slide about augmentation.
Applied AICustomer engagement in B2B sales - the future is agentic
Agentic AI capturing and acting on every B2B sales interaction turns CRMs from static databases into continuous decision engines. If you run revenue operations, the question shifts from "do we use AI for scoring" to "which parts of the sales cycle can we safely let agents execute end-to-end."