Artificial Intelligence as an Investment Case

Artificial intelligence expands what businesses can earn and can accelerate how quickly they earn it. My investment view begins with a rising frontier: exceptional performance becomes harder to sustain, yet the tools, effective research effort, and organizational capacity used to reach the next threshold can improve even faster. The opportunity is not only cheaper existing work. It includes previously uneconomic activities, faster experimentation, new products, and earlier cash flows. This manuscript develops that affirmative thesis across infrastructure, models, applications, scientific discovery, and established operating businesses. It connects customer value to defensible market share, net income, reinvestment, and cash per share. A transparent 2030 reference screen gives approximately USD 5.62 trillion of annual gross AI-related vendor revenue and USD 1.05 trillion of net-income capacity. These are conditional screening quantities, not consolidated final demand or incremental global profit. Dynamic margin compression lowers modeled income to approximately USD 789 billion on the same revenue path; a separate, assumption-dependent capital-renewal overlay gives approximately USD 686 billion. Asset-vintage schedules, matched software and logistics counterfactuals, zero-profit model scenarios, and the value of faster learning expose the mechanisms behind those outcomes. Leopold Aschenbrenner's Situational Awareness supplies a consequential research-automation scenario, not a mandatory timetable or a substitute for financial analysis. The conclusion is constructive and selective: invest in businesses that repeatedly convert improving intelligence into valuable paid activity, retain a differentiated role when intelligence becomes cheaper, and generate cash after financing their next advance.