Why Australian Investor Isn't Betting on Data Centers, AI

Revolution Asset Management Managing Director and Chief Investment Officer Bob Sahota described the firm's conservative approach to private credit, emphasizing cash-flow-based lending. At the Sohn Hearts & Minds Conference in Queenstown, New Zealand, he also explained why the firm is not betting on data centers or selecting AI winners.

Analysis

This is a statement of underwriting philosophy, not evidence of Revolution Asset Management’s portfolio exposure or realized performance; there is no direct public-equity trade from the comments alone. The useful signal is the credit-market fault line: lenders prioritizing current cash generation may be better insulated if AI-related projects face power, construction, tenant-concentration, or refinancing constraints. Conversely, a blanket aversion to data centers could miss assets supported by durable contracts and creditworthy tenants. Cash-flow underwriting is not risk-free either: it can underweight collateral and long-dated growth, while private marks may delay recognition of deterioration. Near term, the comments are unlikely to change asset values; over 1–3 months, monitor fundraising, lending terms, and observable defaults or restructurings for evidence that underwriting standards are diverging. Over 6–18 months, refinancing needs and project-level power availability could separate contracted, operating infrastructure from speculative capacity. The contrarian point is that “AI exposure” is not one risk bucket: project quality, leverage, power access, and tenant credit matter more than the theme label. No exposure, valuation, or performance data are supplied, so do not infer that Revolution is underexposed or that cash-flow-oriented lenders will outperform.

Key Decisions for Investors

• No immediate trade on the interview; treat it as qualitative evidence of a cautious underwriting stance, not a portfolio disclosure or a catalyst.
• For private-credit diligence, compare lender exposure by borrower cash-flow coverage, seniority, covenant protections, maturity profile, and concentration in development-stage versus operating data-center assets.
• Watch for widening differences in private-credit fundraising, loan terms, amendments, and defaults over the next 1–3 months; improving terms for lenders or rising restructurings would challenge the view that underwriting remains disciplined.
• Over 6–18 months, distinguish contracted operating data centers with credible power access from speculative projects; deterioration in tenant commitments, power timelines, or refinancing access would be a thesis-falsifying warning for the broader AI-infrastructure credit narrative.