AI Financing Turns Selective as Lenders Separate Core Borrowers From Riskier Bets

A split in AI credit

Jack Janasiewicz of Natixis told his firm's podcast that capital markets are becoming more discerning in funding AI-related companies, with wider spreads for providers known as neoclouds, and Natixis cut positions in two AI-themed exchange-traded funds at the start of October to move toward direct AI exposure. m:…6399

Barclays analyst Altman, speaking on the Odd Lots podcast, described the market as being in a denial phase about AI trade risks, even as investor consensus across the United States, Europe and Asia remains strongly positive toward AI. m:…2171

The risk, in Altman's account, concentrates in neocloud companies, which sit at the riskier end of AI financing because they lack significant operating cash flow, and a sell-off in CoreWeave bonds followed an attempt by another firm to tap debt markets. m:…2171

Natixis put the scale of the exposure in context, estimating that about 60 percent of S and P 500 market capitalisation relates to the AI theme or its derivatives, which helps explain why a cooling in AI capital spending could shift broader market leadership. m:…6399

What investors still back

Flybridge, in a letter to its limited partners, argued that there is far more value to unlock by building on current AI models, pointing to agents, specialised models and physical AI as its areas of focus. m:…0189

The firm also reported that fewer than one in ten enterprises have AI agents in production, a gap it reads as headroom for applied work rather than as proof that demand has faded. m:…0189

That reading sits alongside striking operating figures cited by Flybridge, including Anthropic revenue of more than 6.6 million dollars per employee on a trailing twelve-month basis and revenue growth of 1,368 percent year over year in the June quarter. m:…0189

Flybridge named the kinds of applied work it is exploring, including a firm helping brands get discovered by agents, a firm helping companies own their own models instead of renting intelligence by the token, and a firm putting autonomous rail cars on the rail network. m:…0189

Barclays survey evidence points the same way on concentration: only 60 percent of investors surveyed believe AI financing will be a problem for the foreseeable future, while about 75 percent of S and P 500 earnings growth next year is expected to come from semiconductors and the broader AI ecosystem. m:…2171

The security gate

Greycroft's survey of security leaders found that 84 percent have had an AI initiative delayed, blocked or cancelled because of security concerns, making security a funding filter as much as a technical checklist. m:…0250

The same survey found that 62 percent of organisations signed at least one early-stage security vendor in the past two years, and that AI-native tooling now accounts for near 10 percent of total security spending, with expectations that the share will double within two years. m:…0250

Greycroft argued that startups win in this market when they are materially better than incumbents, deploy without moving customer data, and fill a gap rather than overlapping with existing spending. m:…0250

The buying process itself is fragmented: in about 40 percent of organisations, the security leader does not hold the budget for AI-native security categories, and only 19 percent of organisations stop evaluating alternatives when an incumbent ships a feature described as good enough. m:…0250

For founders, the combined message is narrow: capital is still available, but it increasingly flows to teams that can show operating cash flow, a security story that survives diligence, and users already moving from pilots into production. m:…2171 m:…0250 m:…0189