BOOMRISK RESEARCH
Evidence-led papers on the spending, financing, demand and competition shaping whether the AI boom pays.
LATEST PAPER
DayOne's billing-based utilisation measures contracted infrastructure demand. Customer compute use and cash recovery after construction and financing remain separate tests.
BOOMRISK RESEARCH LIBRARY
Each paper is tagged by the question it investigates and the pressure tests it informs.
STAY UPDATED ↗Megaport's customers help fund contracted equipment. Replenishing its GPU pool leaves a separate utilisation, funding and cash-return test.
DigitalOcean can spread AI server payments over revenue years. Its public metrics do not yet show a like-for-like cash-flow match.
A broader dashboard can be less informative when related indicators count the same financial pressure more than once.
Strong AI demand does not establish that every layer of the full stack is earning its cost of capital.
Circular financing weakens the independence of some demand signals without proving that AI demand is artificial.
NVIDIA’s latest results strengthen the evidence for leading AI compute demand. The financial returns surrounding the investment cycle remain a separate question.
Every paper records the BoomRisk reading and evidence available when it was published.
Material claims link to primary evidence, with proxy limitations stated explicitly.
Each paper names the pressure tests it informs, keeping analysis tied to the monitor.