The Hangover
Venture capital is climbing out of the deepest trough since the dot-com crash. Global VC funding peaked at roughly $683 billion in 2021, according to CB Insights, fueled by zero interest rates, pandemic-era digital acceleration, and a “growth at all costs” mentality that made profitability seem irrelevant. By 2023, funding had fallen to about $249 billion — a 63% decline. 2024 showed modest recovery to around $280-300 billion. 2025 looks similar: a slow, uneven recovery, not a return to the excess of 2021.
The reset was overdue. The 2021 vintage — startups funded at the peak of the cycle — is going to be one of the worst-performing in VC history. Companies that raised at 100x ARR multiples in 2021 have been marked down, shut down, or sold for fractions of their peak valuations. WeWork’s bankruptcy, FTX’s implosion, and Hopin’s fire sale (from $7.6B valuation to $50M acquisition in three years) are the poster children, but thousands of smaller companies have quietly died.
Dry Powder and the Waiting Game
Paradoxically, there’s still plenty of money. VC dry powder — committed capital that hasn’t been deployed — hit a record $311 billion globally in 2024, according to PitchBook. The money is there; LPs (limited partners — the pension funds, endowments, and family offices that invest in VC funds) are just being more selective about where it goes.
The dynamics that created the overhang: massive fundraising in 2021-2022 when LPs were allocating heavily to venture. Those funds have 10-year lives and need to deploy capital, but at 2024-2025 valuations and pace, deployment is taking longer. The result is a structural supply-demand imbalance: too much capital chasing too few “obviously good” deals, with everyone crowding into AI while ignoring everything else.
Valuation resets have been sharp but uneven. Late-stage startups that haven’t raised since 2021 are carrying valuations that no longer reflect reality — the gap between “paper mark” and “what someone would actually pay” is often 50% or more. Down rounds (raising at a lower valuation than the previous round) went from 3% of deals in 2021 to 18% in 2023, before moderating to about 12% in 2024. Founders who can avoid raising are doing so; those who can’t are accepting painful terms.
Hot Sectors and Where the Money’s Going
AI is eating venture capital. In 2024, AI companies captured roughly 35% of all VC dollars in the US, up from about 15% in 2022. The concentration is extreme: OpenAI ($6.6B round in late 2024 at a $157B valuation), Anthropic ($4B+ from Amazon), xAI ($6B), CoreWeave ($1.1B), and Scale AI ($1B) alone account for a significant fraction of total AI investment. The “AI infrastructure” layer — companies building the picks and shovels for the AI gold rush — has been particularly hot: GPU cloud providers, vector databases, fine-tuning platforms, and LLMOps tools.
Outside of AI, funding has contracted significantly. Fintech funding is down about 70% from 2021 peaks. Enterprise SaaS is down about 60%. Consumer tech is down even more. Healthtech and biotech are relatively resilient, driven by long-term secular trends that don’t depend on ZIRP. Climate tech has been an outlier: funding held up better than most sectors, driven by IRA incentives in the US and genuine technological progress in areas like battery technology, carbon capture, and fusion energy.
The Exit Problem
The VC model depends on exits — IPOs and acquisitions that return capital to LPs. The IPO market was essentially closed from mid-2022 through most of 2024. A few high-profile IPOs in late 2024 (Rubrik, ServiceTitan, Reddit much earlier) have cracked the window open, but the backlog is massive: an estimated 1,200+ VC-backed unicorns globally are waiting to go public.
Many of these unicorns can’t go public at their last private valuation. The alternative — going public at a lower valuation (a “broken IPO”) — destroys morale, triggers anti-dilution provisions for previous investors, and signals failure to the market. The result is a standoff: companies won’t IPO at valuations that clear the market, and public market investors won’t pay 2021 valuations. Something has to give, and it probably will in 2025-2026 as fund lives run out and LPs demand distributions.
M&A as an exit path has been constrained by regulatory pressure. The FTC and DOJ under the Biden administration blocked acquisitions that would have sailed through in previous administrations (Adobe-Figma, Amazon-iRobot, Kroger-Albertsons). This has a chilling effect on M&A, especially in tech, and is particularly damaging for startups whose most likely exit is acquisition by a larger company. The incoming administration may take a different approach, but regulatory uncertainty persists.
What’s Actually Different This Time
The “blitzscaling” era — grow at all costs, profitability later — is over. The companies that survived the post-ZIRP reset have real businesses: growing revenue, positive unit economics, and a path to profitability. The “default dead” companies — burning cash with no plan to stop — are mostly gone. The bar for raising capital in 2025 is the highest it’s been in over a decade.
This is, on balance, healthy. The 2021 bubble misallocated talent and capital into unsustainable businesses. The reset is painful but necessary. The companies being built in 2024-2025 are, on average, more capital-efficient and more fundamentally sound than their 2021 counterparts. Whether that translates into better returns for LPs — the ultimate measure of VC as an asset class — remains to be seen. The 2021 vintage is likely to drag down aggregate VC returns for years. But the 2024-2025 vintage, built on more realistic assumptions and (largely) sustainable unit economics, might actually deliver the returns that 2021 promised.