VCs call the market sheep-like while bond markets feed the AI buildout

The VC critique

VCs Rory O'Driscoll and Jason Lemkin told Harry Stebbings on the 20VC podcast that venture capital is currently broken: investors are ignoring profitable, high-growth software companies while demanding a trillion-dollar market-cap narrative before they will consider funding a company. Capital is concentrating in fewer names and the industry is more sheep-like than ever, they argued, with mega-funds managing $5 billion or more needing billion-dollar positions to make their fund economics work. m:…4305

The numbers they cited sketch the divide. Modal's annual run rate has passed $1 billion after a fundraise at a $15.75 billion valuation; Anduril sits at $90 billion; Anthropic may be priced at $1.5 trillion in a potential IPO; Town raised $108 million at a $1 billion valuation. Meanwhile, software startups targeting specific markets are struggling to secure funding because investors fear AI will render those markets obsolete — and even startups with top-decile growth metrics must now pitch far more VCs than in prior years. m:…4305

The credit wave

While VCs describe a market starving ordinary software of capital, credit markets are feasting. Anish Shah, Rehan Latif, and Samar Oweida of Morgan Stanley told the Thoughts on the Market podcast that credit is seeing a historic wave of issuance driven by AI-related capital expenditure and a resurgence in M&A. Investment-grade debt volumes are up almost 25 percent on the year, they said, with $100 billion of new cash coming into investment-grade credit — and AI-related issuers are relatively rate-insensitive because of optimism about long-term returns on the investment. m:…0871

The Morgan Stanley panel added a caveat: the stress is concentrated lower down the credit stack, with real dispersion and refinancing challenges in the CCC-rated segment. Growth remains the primary driver of credit performance, they said, but corporate derivative business is rising on uncertainty around energy prices, inflation, and fiscal policy. m:…0871

The stress point

BofA Global Research's October fund manager survey suggests enthusiasm is moderating at the margin: 30 percent of respondents now expect a reduction in AI capital spending, more than double last month's reading, and semiconductors as a crowded trade fell to 45 percent from 82 percent in July. The survey also showed the largest technology overweight since August 2020 — positioning is still long, just more nervous. BofA estimates data centers contributed about 70 basis points to U.S. GDP in the first half of 2026. m:…6136

The clearest stress signal is in neocloud debt. Bloomberg's Catherine Schwartz and Dina Bass reported that CoreWeave's stock fell as much as 17 percent last Wednesday and its bonds dipped into distressed territory, reflecting investor skepticism toward the neocloud model: capital-intensive data center builds and GPU purchases create financing hurdles for companies without hyperscaler balance sheets. One co-founder said CoreWeave is exploring investment-grade borrowing, but investors say a neocloud deal clearing the market right now would be difficult or prohibitively expensive — even though CoreWeave and most public neoclouds are sold out of capacity. NVIDIA invested $2 billion in CoreWeave earlier this year, and Bloomberg notes CEO Jensen Huang defends the company because NVIDIA needs its chip customers solvent to meet its own growth forecasts. Volta Infrastructure recently priced a $5 billion leveraged loan at what were described as among the highest rates paid to finance the AI buildout, and UK-based Nscale faces pressure to lower its IPO valuation range. m:…9733

What it means

This is my read, not a reported fact: two capital markets are disagreeing. Private risk capital is concentrating into fewer, larger AI bets while public credit eagerly funds the buildout — and neoclouds sit at the fault line, sold out on demand but fragile on the balance sheet. The hyperscalers' financing advantage (Microsoft, Amazon, Google can fund from their own balance sheets) may widen into a structural moat if neocloud refinancing stays expensive. For a VC, the question is not whether AI capex continues, but who gets to finance it — and what that implies for startup leverage in the next round. m:…9733 m:…4305