For the past three years, the AI investment cycle has largely rewarded speed.

More capable models attracted more capital. That capital financed expanding computing capacity, semiconductor production, data centres and energy infrastructure. Expectations of exponential growth supported rising valuations across public and private markets. However, developments in September and October 2026 suggest that the investment landscape is changing. AI safety concerns, rising infrastructure financing requirements and the approaching IPOs of leading AI developers are bringing greater scrutiny to the economics behind the industry’s expansion.

The central question is shifting from how quickly AI can advance to whether its growth can generate sustainable returns on invested capital.

For private equity and venture capital investors, this transition has significant implications.

AI Safety Meets Competitive Reality

In September, Anthropic CEO Dario Amodei called for a more deliberate pace of frontier AI development, warning that capabilities are advancing faster than the industry’s ability to understand and control their risks.

OpenAI CEO Sam Altman subsequently supported greater caution and ruled out an OpenAI IPO in 2026, citing safety concerns. Yet slowing development presents a competitive dilemma.

Anthropic continued releasing new models amid growing competition from OpenAI, while the US and China remain committed to technological leadership. This introduces pacing risk into the AI investment equation. Stricter safety requirements could affect development timelines and infrastructure spending, although slower frontier development would not necessarily reduce demand for existing AI applications.

Competition continues to favour acceleration, even as safety concerns become increasingly important.

Revenue Growth Faces the Profitability Test

Anthropic’s IPO preparations illustrate the industry’s economic challenges.

According to Reuters, Anthropic generated approximately $4.6 billion in 2025 revenue, representing twelvefold annual growth. However, it recorded an operating loss exceeding $8 billion and spent $7.33 billion on computing and infrastructure. Its prospectus also outlines approximately $518 billion in future cloud, computing and infrastructure obligations. These figures illustrate the capital required to maintain technological leadership.

Meanwhile, competitive pressure on inference prices raises questions about long-term margins. The challenge is that rapid revenue growth does not automatically translate into sustainable profitability. This distinction became particularly visible when OpenAI’s annualised September revenue was reported at approximately $50 billion, compared with an earlier indication approaching $70 billion. The discrepancy primarily reflected differences in revenue reporting rather than deteriorating customer demand. Nevertheless, the market reaction highlighted growing sensitivity to AI financial expectations.

Investors must increasingly distinguish between commercial adoption, reported revenue, profitability and free cash flow.

Infrastructure Financing Becomes a Critical Constraint

The same pressure extends throughout the AI value chain.

On October 12, Broadcom CEO Hock Tan identified capital availability as a major constraint on AI expansion. A proposed $60 billion financing package involving Broadcom and Anthropic illustrates the industry’s increasing dependence on institutional capital. The structure reportedly includes $42 billion in senior secured financing and $18 billion in junior debt. M

eanwhile, Financial Times reporting shows AI-related debt issuance declining to approximately $23 billion in September, following a June peak of $113 billion. Although earlier borrowing partly explains the slowdown, financing conditions are becoming an increasingly important consideration. For lenders and infrastructure investors, the question is whether future AI demand can justify capital expenditure, debt obligations and long-term infrastructure commitments.

Counterparty concentration, asset utilisation, technological obsolescence and refinancing risk are therefore becoming central underwriting considerations.

Public Markets Will Test Private Valuations

These developments are increasingly influencing public-market expectations.

Although leading frontier AI developers remain privately held, their spending supports revenues across listed semiconductor manufacturers, cloud providers and infrastructure companies. Changing expectations for private AI companies can therefore affect public valuations throughout the supply chain. The anticipated IPOs of Anthropic and OpenAI could introduce more direct public-market price discovery, exposing their profitability, capital intensity and valuation assumptions to broader investor scrutiny. Recent developments suggest that ambitious private valuations will not necessarily translate into successful listings.

In October, Nvidia-backed AI infrastructure company Firmus Technologies withdrew its planned Australian IPO, which had targeted a valuation exceeding $30 billion, approximately three times its valuation two months earlier. The withdrawal illustrates the difference between valuations negotiated in private funding rounds and prices public investors are willing to support.

This reinforces the importance of continuously reassessing valuation assumptions. This reinforces the importance of regularly reassessing private-market valuations against relevant benchmarks. Sector-, stage- and region-specific EV/Sales and EV/EBITDA multiples, as reflected in Venionaire DealMatrix Multiples, can provide additional context as market conditions and investor expectations evolve.

Venture Capital: Capital Concentration and Exit Discipline

Despite growing scrutiny, venture investment remains substantial. However, capital is increasingly concentrated in a limited number of companies and larger financing rounds.

Venionaire Capital’s European Venture Sentiment Index (EVSI) for Q3 2026 rose to 5.4, exceeding the expected 4.4 and signalling improving investor confidence. Yet 56.2% of the $24.7 billion invested across European venture capital went into late-stage rounds, while seed and angel financing accounted for just 7.4% of capital, highlighting increasingly selective capital allocation.

For VC investors, upcoming AI IPOs could provide liquidity and establish clearer valuation benchmarks. Conversely, disappointing public-market valuations could pressure late-stage portfolio marks and exit expectations. Investment strategies must also distinguish between capital-intensive frontier developers and AI application companies with potentially more scalable economics. For application companies, declining model costs can improve margins while simultaneously lowering barriers to entry.

Defensibility becomes essential. Proprietary data, customer relationships, distribution, workflow integration and measurable productivity benefits may provide more sustainable competitive advantages than access to AI technology alone.

Private Equity: The Productivity Gap

For private equity, AI presents opportunities for operational improvement, but the financial benefits are far from automatic.

Gerhard Kürner’s September 2026 report, Work After AI, highlights the gap between individual productivity improvements and measurable company-level results. Drawing on McKinsey research, the report finds that 80% report individual productivity gains, while only 37% identify benefits in operating results. Just 6% report a measurable operating-result improvement of at least 5%.

This distinction matters for PE investors.

AI adoption may increase employee productivity without necessarily improving EBITDA. Capturing financial value often requires organisational restructuring, workflow redesign and changes in resource allocation rather than simply deploying AI tools. At the same time, AI threatens established business models, particularly in software and labour-intensive services.

According to the Bank for International Settlements, outstanding private credit loans to software businesses exceeded $500 billion by the end of 2025, representing approximately 19% of direct lending. AI-driven disruption could therefore affect both equity valuations and borrowers’ creditworthiness.

For PE investors, acquisition underwriting must increasingly assess whether historical earnings, recurring revenues and competitive advantages remain defensible. The opportunity lies in identifying businesses capable of translating AI adoption into measurable operational improvements while maintaining sustainable competitive positions.

From Acceleration to Validation

The AI investment cycle is not necessarily ending. Demand continues expanding, technological progress remains rapid and capital deployment remains substantial.

However, technological transformation does not guarantee attractive investment returns.

For public markets, the next phase will test profitability and infrastructure economics. For venture capital, it will test private valuations, liquidity and business model defensibility. For private equity and private credit, attention will increasingly shift toward operational resilience, cash-flow quality and disciplined underwriting.

After years dominated by acceleration, the next phase of AI investing may be defined by financial validation: the ability to translate technological progress into durable economic returns.

Sources