The AI Questions That Matter Most

AI is driving a capital-intensive industrial renaissance in which infrastructure bottlenecks, nuanced disruption and management quality will determine the investment winners.

• Apollo estimates that data centers, power infrastructure, chips and networking equipment could require $5 trillion to $6 trillion of investment through 2030, exceeding the capacity of traditional venture capital, free cash flow and public equity markets.
• The current buildout differs from telecom’s overcapacity cycle because many projects have identified customers with long-term commitments, while power, labor, permitting, transformers, semiconductors and grid connections constrain supply.
• Cloud adoption has taken roughly 20 years since Amazon launched AWS in 2006, with only about half of compute workloads estimated to have moved to the cloud; AI infrastructure is expanding much faster, with cumulative AI-related capex projected at $925 billion in 2026, $1.975 trillion in 2027 and $5.625 trillion in 2030.
• Apollo’s thematic-investing team evaluates software disruption by workflow susceptibility to large language models alongside switching costs, regulatory barriers, proprietary data, network effects and the cost of failure; it expects disruption to spread as foundation models improve.
• The impact on labor and software will not be binary: AI may automate tasks and displace some jobs while creating new professions and expanding demand, and software companies will diverge according to their ability to adapt.

Takeaway: Investors should focus on the infrastructure financing opportunity and on software companies with strong defenses and management teams that adapt quickly, while recognizing that the projections are forward-looking and outcomes may differ materially.