Capital & Industrial Strategy

OpenAI and Anthropic are both moving toward public listings while grappling with a structural challenge: compute costs that scale faster than revenue. The Wall Street Journal reports both companies are burning capital at rates that could exceed $5 billion annually, driven by training runs, inference costs, and the need to maintain competitive model performance. Neither has disclosed a path to sustainable margins at current pricing levels. OpenAI's revenue is growing rapidly but remains dwarfed by Azure compute commitments, while Anthropic's enterprise contracts are longer-cycle and haven't yet offset its capital intensity. Both firms are likely to pitch growth narratives rather than profitability timelines to public market investors.

The IPO preparation comes as both companies face pressure to justify valuations set in private rounds during 2024-2025. OpenAI's last private valuation reportedly exceeded $150 billion; Anthropic's was in the $30-40 billion range. Public market scrutiny will focus on customer concentration risk, gross margins, and whether enterprise adoption can offset consumer product subsidies. Neither firm has published audited financials, and the path to profitability remains speculative.

Private equity buyouts fell 36% quarter-on-quarter to $172 billion in Q1 2026, the sharpest contraction in over two years, according to the Financial Times. The pullback reflects two forces: geopolitical uncertainty from Middle East conflict and, more structurally, investor caution around businesses vulnerable to AI-driven margin compression or workflow automation. PE firms are struggling to underwrite exit multiples for traditional service businesses, professional services firms, and back-office-heavy enterprises where AI could eliminate headcount or commoditise offerings within the typical 5-7 year hold period.

Capital is reallocating rather than sitting idle. The same report notes that infrastructure and technology buyouts held up better, and alternative data suggests private credit and direct lending into AI infrastructure projects grew sharply in the same quarter. The implication is that PE is bifurcating: legacy buyout strategies are stalling, while growth equity and infrastructure funds are leaning into AI enablement.

AI is now being deployed in auditing, compliance, and identity management workflows at scale, but regulators have not yet issued sector-specific guidance. The Financial Times reports that AI tools are already being used to review company accounts, flag anomalies, and automate control testing in financial audits, raising questions about liability when models miss material misstatements or generate false positives. Okta CEO Todd McKinnon told Semafor that AI agents in enterprise identity systems create new attack surfaces, particularly around privilege escalation and session hijacking, which existing cybersecurity frameworks do not address. A separate Bloomberg report highlighted a startup offering AI agents that join every Zoom call and nudge employees to complete tasks, raising unresolved questions about employee surveillance, consent, and data retention under existing labor and privacy laws.

Regulatory agencies including the SEC, PCAOB, and state labor boards have not issued formal guidance on AI use in these contexts. Firms are moving ahead with deployments based on internal risk assessments, creating a patchwork of practices that will likely face retrospective scrutiny when the first material failure occurs.

Hon Hai Precision Industry, Nvidia's largest contract manufacturer, reported a 29.7% rise in quarterly sales driven by sustained AI hardware demand, according to Bloomberg. The growth occurred during the early weeks of Middle East conflict, suggesting that enterprise AI infrastructure buildouts are continuing despite macro uncertainty. Hon Hai's results indicate that hyperscaler capex plans for 2026 remain intact and that supply chain diversification efforts outside China are proceeding faster than anticipated.