General Catalyst (GC) just published its first-ever quarterly investor letter, and the headline figure is not the $380 billion valuation Anthropic fetched in its Series G; it is the state of an entire asset class. CEO Hemant Taneja argues that software buyouts structured around terminal value, rather than cash flow, are now delivering negative returns even when the underlying business performs. The firm is betting $43 billion in AUM that the next era of private markets looks nothing like the last.
The quarterly review, signed by Taneja, and published April 21, lands at a moment of maximum stress for growth-stage software investors. Public software multiples have compressed to roughly 12.7x EBITDA, compared to the 25x entry multiples that characterized 2019–2021 buyouts, destroying equity value even in businesses that grew EBITDA by 50%. Taneja runs the math explicitly: a hypothetical deal entering at 25x EBITDA, with the business performing, still returns 0.68x MOIC and a negative 7% IRR when multiples revert to current levels. The business succeeded but the investment model failed.
The GC portfolio is not protected from this dynamic, but Taneja is positioning it as the exception. The firm participated in Anthropic’s Series G, one of the largest venture rounds in history, at a $380 billion valuation, up from the $61.5 billion at which GC first invested in March 2025. That 6x markup in twelve months makes the round one of the best-performing venture bets in the asset class. Anthropic’s annualized revenue run rate grew from $1 billion to over $20 billion in fifteen months, a scaling trajectory GC acknowledges has no precedent.
More structurally significant than any single investment is what GC calls its Customer Value Fund (CVF), which this quarter became, by the firm’s own account , the first venture vehicle ever to achieve an investment-grade credit rating. BBB or higher now covers roughly two-thirds of CVF’s capital. Apollo Global Management which oversees nearly $800 billion in investment-grade credit strategies that venture has historically been unable to touch, because LP structures carry no defined duration. CVF was designed to bridge that gap, and the rating means GC can now bring that capital to bear for its companies without betting on exit multiples.
The geographic ambition is equally large. In February, GC announced a $5 billion commitment to India over five years, which Taneja frames as one of the largest single-country investment commitments ever made by a venture firm. The logic is compounding: India’s UPI now processes more transactions than Visa and Mastercard combined; Aadhaar is the world’s largest biometric identity system; and one million young people enter the workforce every month. The constraint, per Taneja, is not infrastructure or capital, it is the absence of AI-native companies built to leverage both.
The SaaS disruption thesis running through the letter is a major bet. GC cites a portfolio company that eliminated five SaaS providers by rebuilding functionality in-house, saving $8 million annually. Taneja draws a line between two categories of software that will survive the transition: point solutions built around headcount growth, which are structurally impaired; and deeply embedded platforms sitting on proprietary operational data, the healthcare system with ten million patient cases, the financial institution with full-cycle default data, where AI amplifies the moat rather than commoditizing it. Stripe’s $1.9 trillion in 2025 payment volume is the archetype.
Inside GC, an operating company called Percepta is doing applied AI research that the letter publishes in full. The team’s Q1 finding: a WebAssembly interpreter can be implemented inside transformer weights, enabling a language model to execute arbitrary C code internally rather than delegating to an external tool. The key technical unlock is a decoding path that turns attention lookups from linear scans into logarithmic-time queries, generating execution traces at more than 30,000 tokens per second on a CPU. The system solved Arto Inkala’s Sudoku, widely cited as the hardest in the world, in under three minutes. The implication for enterprise AI deployment, per GC, is a path toward systems that can both reason flexibly and compute reliably inside a single inference loop.
Taneja’s most operationally unusual bet remains the acquisition of Summa Health, a hospital in Akron, Ohio; the first time a venture capital firm has purchased a hospital outright. The move is framed as partnership: the firm hired former healthcare CEOs including Ken Frazier and Kate Walsh to run alongside tech-native investors. Taneja argues explicitly that legacy institutions are not obstacles but partners, and that the industries most in need of transformation are governed by people with decades of knowledge that venture does not have.
For investors and founders, the GC letter maps a interesting, new VC blueprint. The buyout firms that entered software at 25 to 50x free cash flow are facing a category repriced underneath them, with no obvious exit: strategic acquirers are managing their own AI transitions, and the public markets for sub-$10 billion companies have effectively closed. The firms that entered at 10x on a cash-flow thesis, or that have built vehicles like CVF to generate returns without terminal value dependency, is advantaged. The question for every software investor is now Taneja’s: if software is abundant, what is actually scarce? His answer: institutional trust, deep customer relationships, operational knowledge that compounds with increasing cognition, is not an abstraction. It is the investment thesis GC is betting $43 billion on.
Read the full article here
