Venture Capital Trends 2026: Where the Money Is Moving and Why

TL;DR: Global VC hit roughly $285B in 2025, a recovery from the 2023 low but still far below the 2021 peak of $621B. AI took an estimated 34% of all investment, with defense and climate tech also drawing outsized interest. This is a reset, not a rebound: fewer and larger checks, lower valuations, longer holds, and capital flowing to investors who prize discipline over speed.

---

Global venture capital investment reached about $285 billion in 2025. That is a real recovery from the 2023 low of $238 billion, but it sits far below the 2021 peak of $621 billion. The headline number matters less than where the money went. AI took the largest share by a wide margin. Climate and defense technology pulled in more institutional money than their history would predict. The rest of the market looks less like a rebound and more like a reset, with capital still flowing but flowing on different terms than it did when money was free. Managers who raised record funds in 2020 and 2021 are now deploying into a market where discipline counts for more than speed.

Key Takeaways

• Global VC investment hit roughly $285B in 2025, up from $238B in 2023 but below the $621B peak in 2021.
• AI and AI-infrastructure companies took an estimated 34% of all global VC investment in 2025.
• Median seed rounds grew from $2.1M in 2021 to $3.4M in 2025. Median Series A rounds shrank slightly, from $18.5M to $16.2M.
• Defense technology investment reached $35B globally in 2025, more than double its 2022 level.
• Climate tech stayed above $60B for the third year running.
• Down rounds among late-stage companies peaked at 31% of deals in 2024, then fell to 19% in 2025 as valuations settled.
• U.S. deal count dropped 18% from 2022 to 2024, with money concentrating into fewer, larger checks.

Investment Volume, Stage, and Sector

The funnel has changed shape since 2021. Seed activity has held steady because early-stage software is still cheap to test. Series A and Series B counts fell harder, a sign of pickier investors and the backlog of 2021-vintage companies still working through their cash. So the gap between companies that raise a seed and companies that graduate to an institutional Series A has widened. Seed-to-Series-A conversion dropped from around 30% for 2020-vintage cohorts to closer to 20% for 2023-vintage cohorts. Both a higher bar at Series A and a much larger pool of seed-funded companies feed that decline.

Early rounds inflated even as later stages tightened. The jump in median seed size from $2.1M to $3.4M reflects two things: institutional funds showing up earlier, and the higher cost of building in AI. GPU compute, data, and technical salaries pushed the floor up for anything AI-adjacent. Seed sizes for non-AI startups stayed flatter, around $2.5M.

AI’s 34% share is a level of concentration the market has not seen in a single sector since the mobile and internet-infrastructure waves of earlier decades. Model infrastructure, developer tools, and vertical AI applications are at once the most defensible and the most crowded places to invest. About half of that capital went to the infrastructure layer: compute providers, training platforms, and foundation-model developers. The rest split between developer tooling, which drew heavy early-stage interest, and vertical applications across healthcare, legal, financial services, and enterprise workflows. The vertical layer is where competition is fiercest, with hundreds of companies building similar products on the same handful of models. Investors are betting that distribution and proprietary data, not raw model quality, will pick the winners.

Defense technology’s climb to $35B marks a genuine shift. A decade ago the category was off-limits for many institutional funds. Geopolitics since 2022 changed that, and defense tech now sits in most major portfolios. Autonomous systems, cybersecurity, satellite intelligence, and communications lead the way. The move is not just American. European defense tech accelerated as the EU raised its spending targets, and Israel, Australia, and South Korea all saw more defense-adjacent activity. Talent followed the money. Engineers who once headed for consumer internet or fintech are joining defense startups at the highest rate since the early 2000s.

Climate tech holding above $60B a year stands out because it survived the broader correction. Federal policy supplies non-dilutive money alongside equity, which lowers the risk on each deal. Tax credits from the Inflation Reduction Act, Department of Energy loan guarantees, and similar tools in the EU and UK have built a financing stack that earlier climate startups never had. Battery storage, grid infrastructure, carbon capture, and industrial decarbonization see the most volume. Failure rates here run higher than in software because the work is capital-heavy and slow, but the blended public-private model has helped more companies survive once they get past first commercialization.

Biotech and life sciences get less attention in an AI-dominated cycle, but they still drew roughly $38B globally in 2025. The sector runs on its own clock, set by clinical readouts and regulatory milestones rather than hype. AI-driven drug discovery created an overlap that pulled in extra interest, with computational biology companies raising on a mix of biotech fundamentals and AI-premium valuations.

Valuations, Fund Economics, and Returns

Funds from the 2018 and 2019 vintages are posting solid IRR, in the 18% to 24% range for top-quartile managers. The 2021 vintage is under pressure, and early marks suggest the high entry prices of that period will eat into realized returns. The spread between top-quartile and median performance has widened. For 2018 and 2019 vintages, the gap in net IRR runs about 14 points. For 2021 it is likely to be wider still, as the managers who held their pricing discipline pull away from the ones who paid the peak.

Median pre-money at seed fell from about $12M in 2021 to $9.5M in 2025. Series A pre-money dropped from $65M to $42M, a 35% compression that gave investors better entry points. Series B and later fell harder in some categories, with median Series B pre-money sliding from $180M to roughly $115M. AI partly dodged the reset: median AI Series A pre-money sat near $55M in 2025, well above the cross-sector figure.

Down rounds peaked at 31% of late-stage deals in 2024. The drop to 19% in 2025 came from a mix of real recovery and the clearing of forced 2021-vintage repricings. Flat rounds added another 22% of late-stage deals in 2025. Put those together and more than 40% of late-stage deals last year were flat or down, which says valuations are still normalizing even as the headline down-round number improved.

Venture debt grew as a companion to equity. Among surveyed venture-backed companies, 34% have used it, up from 21% in 2021. Higher equity costs and a more mature lending market both played a part. Silicon Valley Bank’s 2023 collapse disrupted the market at first, but HSBC Innovation Banking, Trinity Capital, and Western Technology Investment expanded to fill the gap. Revenue-based financing and other non-dilutive options also gained ground, especially among SaaS companies with predictable recurring revenue.

Exits take longer now. Median time from Series A to exit ran about 8.2 years for companies that exited in 2025, up from 6.4 years in 2019. Longer holds drag on IRR even when the final multiple holds up. GPs have responded by extending fund terms, raising continuation vehicles, and doing more GP-led secondaries to manage timing.

Where the Money Is Geographically

The U.S. held about 48% of global deal value in 2025, down from 53% in 2021. Both the growth of other ecosystems and the disappearance of 2021’s mega-rounds explain the slip. The San Francisco Bay Area alone took roughly 30% of all U.S. investment by value, almost entirely on the back of AI headquarters and infrastructure rounds. New York stayed the second-largest hub, strong in fintech, media tech, and enterprise SaaS. Secondary hubs softened, with Austin, Miami, and Denver seeing slower deal growth after the 2020-to-2022 relocation surge.

India drew about $18B in 2025. The market has grown past consumer internet and fintech into enterprise software, SaaS, and AI-enabled services. The IPO window opened up, and several high-profile listings in 2024 and 2025 gave the ecosystem the exit proof it had been waiting for. Bangalore is still the main hub, with Hyderabad and Pune building secondary clusters in enterprise and deep tech. Stable rules for foreign investment kept international LPs interested.

Europe recovered to roughly $48B in 2025, with AI and deep tech taking an outsized slice. The UK, France, and Germany account for about 70% of European deal value. London’s AI cluster drew global attention. France built a strong research-to-startup pipeline backed by government programs and university spinouts. The Nordics keep producing more startups per capita than anywhere else, led by fintech and climate. Europe still struggles at the growth stage, where companies often turn to U.S. investors to lead the larger rounds.

China-facing investment by U.S. managers kept falling, down to an estimated $8B in 2025. Regulation, geopolitics, and LP pressure all pushed U.S. money out of Chinese markets. Domestic Chinese VC, funded mostly in RMB, stays active but has shifted toward semiconductors, advanced manufacturing, and strategic technology that lines up with government priorities. The split between the two ecosystems is now a permanent feature, not a passing disruption.

Southeast Asia drew about $9B in 2025, led by Indonesia and Singapore. Activity cooled from the 2022 peak but holds a base in fintech, logistics, and digital commerce. Latin America, the Middle East, and Africa together came to about $12B, each region building more local funds and a more sophisticated startup base.

The Major Firms

Andreessen Horowitz (a16z) has become the most visible institutional firm through an AI-first thesis, a media operation, and accelerator programs. Organizing its teams around sector practices, including crypto, bio, games, and infrastructure, gives it a depth that generalist funds lack.

Sequoia Capital runs global coverage across the U.S., Europe, India, and Southeast Asia. Its move to an open-ended fund structure, and the later reversal, was one of the most-watched structural experiments in the industry.

Lightspeed Venture Partners focuses on enterprise SaaS, consumer tech, and emerging-market growth, with real presence in India and Southeast Asia alongside its U.S. work.

General Catalyst added health-system transformation and AI to its traditional SaaS and consumer practice. Its operating partnerships with health systems put it on the hook for operational risk, not just capital, which is unusual in venture.

Coatue Management works across public and private markets with heavy technology exposure, which gives it a valuation read that pure venture firms do not have.

Founders Fund runs a deep-tech and defense focus, with major bets in space, AI, and biotech. It will back capital-heavy, long-duration companies that faster software-focused funds avoid.

Khosla Ventures invests in deep tech and AI across climate, health AI, and semiconductor design, and has one of the longer track records in climate specifically, well before the current wave.

Bessemer Venture Partners keeps a SaaS-focused practice with one of the longest records in enterprise cloud, and its cloud index is a widely cited benchmark for public SaaS performance.

Tiger Global cut its private deployment sharply from the 2021 peak but is still a relevant late-stage investor. Its retreat from high-volume, low-diligence dealmaking reflects the broader change in crossover behavior since 2022.

GV (Google Ventures) offers corporate exposure to enterprise software, life sciences, and AI, with access to the Google ecosystem. It operates more independently than most corporate funds, though portfolio companies still benefit from Google’s infrastructure and reach.

How the Models Differ

Traditional equity funds and rolling funds differ on LP liquidity, capital concentration, and speed. Rolling funds raise continuously and move fast; traditional funds concentrate capital in defined vintages with fixed investment periods. Rolling funds caught on from 2020 to 2022, especially among emerging managers, but pulled back as LPs favored the accountability of the traditional structure. Total capital in rolling vehicles is still small, under $5B globally.

Generalist and specialist funds diverge by sector. When one sector runs hot, specialists beat generalists inside their focus area. Generalists offer better coverage across cycles. The AI cycle has rewarded specialist AI funds, but history says heavy sector concentration carries real risk when the cycle turns. The best performers over decades tend to combine deep sector knowledge with the freedom to move as opportunities shift.

Corporate venture plays its own role, deploying with strategic goals that differ from pure return-seeking. CVC deal count grew 11% in 2024 as large tech and industrial companies used minority stakes to reach AI, climate, and sector software. That CVC programs held up through the correction is notable, since past downturns shut many of them down. This cohort looks more committed, with dedicated funds, independent decisions, and clearer mandates than the corporate programs of the early 2010s.

Angel and syndicate platforms remain the on-ramp for individual investors and smaller institutions. AngelList, Republic, and similar services professionalized angel investing and added data transparency that early-stage markets never had. The capital is modest next to institutional VC, but their role in price discovery and deal sourcing at pre-seed and seed is real.

The Outlook Into 2027

AI concentration will hold through 2027, with a growing split between infrastructure, which keeps drawing capital, and applications, where returns will land with a small number of winners. The application layer faces real compression as companies built on the same models compete on execution and distribution rather than technology. Infrastructure has higher barriers and more defensible positions, so it will keep pulling growth-stage money.

Exit conditions should improve as IPO windows reopen. Rate stability, a public-market recovery, and time clearing the 2021-vintage overhang set up more IPO activity in 2026 and 2027. The backlog is large. More than 300 U.S. venture-backed companies meet the traditional thresholds for going public: $100M or more in annual recurring revenue and operating margins at or near positive. The pace depends on public-market appetite and the Fed’s rate path, but the supply side of the pipeline is full.

M&A will rise as a second exit channel. Big tech companies with strong balance sheets and an appetite for AI capability will buy venture-backed companies, especially in the $500M to $5B range. Regulatory scrutiny of large deals stays a factor, but transactions below the mega-deal line draw less review.

Secondary liquidity will keep growing as a permanent feature of the asset class. The $130B secondary market of 2025 is expected to top $200B by 2027 as LP-driven restructurings and GP-led continuation vehicles expand. The market is professionalizing fast, with dedicated funds raising larger vehicles and pricing getting tighter. That liquidity layer changes the risk picture for LPs, who used to face 10-to-12-year lockups with little interim liquidity.

Fundraising stays hard for emerging managers. LPs are concentrating with established firms that have track records through multiple cycles. First-time fund sizes have shrunk, and fundraising timelines have stretched past a year in many cases. The new managers who break through will have differentiated sector access, proprietary deal flow in high-demand areas like AI and defense, and operational value beyond the check.

Methodology

This report draws on aggregated venture investment databases, public fund-performance research, LP reporting from pension and endowment transparency rules, and modeled volume estimates where comprehensive public data is missing. Stage classifications follow standard industry definitions: seed covers pre-seed through seed extension, and Series A, B, and later are classified by round label as reported in transaction databases. Geography is based on company headquarters. Sector is based on primary business activity at the time of investment. All figures are in USD unless noted.

The Bottom Line

Venture capital in 2026 is more disciplined and more selective than it was at the 2021 peak, and far more concentrated in AI. The reset has produced better entry points, tighter capital deployment, and clearer priorities. AI, climate, and defense are the thematic story of the decade’s back half, and the investors who moved into them fro