Startup and Venture Capital News — Wednesday, 29 July 2026: Record $510 Billion, AI Capital Concentration, and Open IPO Window

The venture capital market as of late July 2026 is in a state that is difficult to describe in a single word. Formally, this is the best year in the industry’s history: global venture investments in the first half reached a record $510 billion, exceeding the entire 2025 total ($440 billion) and the previous half-year peak of the second half of 2021 by about one-third. In reality, however, investors are facing a market of extreme concentration, where nearly half of all capital goes to two companies, while the number of deals is not growing. For venture funds and institutional investors, the key question in July is not “is there money?” but “who gets it and on what terms.”

Key figures for 29 July 2026: the numbers shaping the agenda

Below are the reference metrics around which the current market discussion revolves:

$510 billion – global venture investments in the first half of 2026; Q1 delivered $305 billion, Q2 another $205 billion across more than 5,000 companies.
43% – the share of two companies, OpenAI and Anthropic, in global venture funding for the half-year (a combined $217 billion).
Over 70% – the proportion of AI-related startups in global venture investments in Q2, compared with roughly 50% a year earlier.
$412.7 billion – venture investments in the US over the half-year, of which $355.9 billion (86%) went to AI companies.
$251 billion – raised in 86 US IPOs since the start of the year, more than five times the full-year 2025 total ($47.4 billion).
$113 billion – the volume of startup acquisitions valued at $1 billion or more in Q2, a record on record.
5.09 billion rubles – the volume of the Russian venture market over the half-year, down 40% year-on-year, with the number of deals halved.

A half-year record: why $510 billion does not mean “the market is back”

The record venture funding volume was not achieved through a broader funnel but through a handful of mega-rounds. The number of deals in the first half was virtually flat, and in Asian markets transaction volumes fell to multi-year lows despite record sums. In other words, the average cheque has increased multiple times while access to capital has narrowed.

Late-stage financing in Q2 rose roughly 141% year-on-year. This is a fundamental shift in venture fund behaviour: capital is flowing not into expanding portfolios with new names but into recapitalising proven leaders. For fund managers, this means a more predictable but less asymmetric return profile; for LPs, it means a growing correlation between funds pursuing different strategies.

Capital concentration: the key risk on the agenda

A situation where two companies absorb 43% of global venture capital over a half-year is historically unprecedented. Add to that a geographical skew: around 88% of all AI-startup investments go to US-headquartered companies. At the same time, the US share of total Q2 volume fell from 83% to 66–67% – capital is simultaneously concentrating by sector and internationalising by geography.

For investment committees, this raises three practical questions:

How diversified is a fund’s portfolio if the bulk of sector returns are driven by a handful of private companies?
How should “second-tier” AI startups be valued when valuation benchmarks are set by rounds of unprecedented scale?
What happens to sector multiples if at least one of the leaders disappoints the public market?

Deals in late July: where the money actually went

The last ten days of July provided a telling snapshot of venture fund priorities. The most notable funding rounds include:

Etched – $300 million, Series C, inference chips, led by Sequoia.
CuspAI – $450 million, Series B, AI for new materials discovery (Kleiner Perkins, NEA).
Meshy – about $400 million, Series B, 3D content generation.
Glow – $180 million, Series A, cybersecurity (Sequoia, Cyberstarts).
Cathedral – $160 million, defence cyber-AI (Andreessen Horowitz, Sequoia).
Humanoid – $152 million, Series A at a $1.35 billion valuation; the first European “unicorn” in humanoid robotics.
Neo – $100 million upon exiting stealth mode, application security in the age of AI agents.

Earlier in July, the market saw even larger transactions: $1.8 billion for defence-focused Helsing, $1 billion for SambaNova Systems, $800 million for Together AI, $700 million for medtech platform Neko, and €411 million for fusion project Proxima Fusion. The overarching takeaway: venture capital is funding not so much applications as the “operating system” of the new economy – compute, energy, security, and industrial robotics loops.

Physical AI, defence, and deep tech: a new map of priorities

Three themes are shaping investment sentiment in the second half of 2026. The first is physical AI: models combined with hardware, from construction robots to industrial perception. The second is defence and sovereign technologies, where European startups are competing with their US counterparts in cheque sizes for the first time in a decade. The third is energy for data centres: fusion, geothermal, and grid projects are being financed as an infrastructure rather than venture asset class.

It is also telling that cybersecurity has evolved into a derivative of AI agent proliferation: investors are funding companies that solve problems created by generative models themselves. This is a stable “second-order” pattern and will remain a source of deals at least until year-end.

IPO window 2026: open, but not for everyone

The primary market is experiencing its strongest comeback since 2021. By late July, the US had seen 86 IPOs with a combined volume of $251 billion; global proceeds for the half-year reached $178 billion (+205% year-on-year) across 524 deals. Tech listings averaged a 44.5% first-day pop, and the combined valuation of companies in the IPO pipeline exceeded $2.1 trillion.

Yet the structure of this record is as concentrated as the venture market. SpaceX’s $85.7 billion listing at a $1.75 trillion valuation accounted for roughly one-third of all funds raised for the year. Anthropic filed on 1 June after a $65 billion round, and OpenAI filed confidentially on 8 June at a private valuation of $852 billion. Strava is preparing a listing at around $2.2 billion. Meanwhile, Databricks publicly declined a 2026 listing in favour of 2027, discussing a private round at a $165–175 billion valuation versus $134 billion six months earlier. Canva and Cohere are still viewed by the market as 2027 candidates.

M&A and exits: the best quarter in five years

For the first time since 2021, exit dynamics have caught up with funding dynamics. In Q2, 32 companies went public with valuations above $1 billion, and another 24 were acquired at prices of $1 billion or more, for a total of $113 billion – a record on record. For venture funds, this means a release of DPI: LP distributions are finally returning to levels that allow a full cycle of new fund subscriptions.

Nevertheless, exit quality remains uneven. Large strategic acquisitions are concentrated in AI infrastructure, semiconductors, and biotech, while classic mid-sized SaaS is still exiting at a discount to 2021 rounds.

Fundraising and dry powder: capital exists, but access is limited

Globally, private markets hold around $3.9 trillion in undeployed capital, of which roughly $600 billion is directly in venture funds. At the same time, the share of successfully closed funds has fallen to about 57%, compared with 94% in 2020 – LPs have become markedly more selective and prefer proven platforms over new managers.

The practical consequence for the market: the gap between top-quartile and other funds continues to widen, and emerging managers are increasingly entering deals through syndicates, SPVs, and co-investments with large platforms.

Russia and the CIS: a market in hard-selection mode

The Russian venture market is moving in the opposite direction to the global trend. In the first half of 2026, venture investments totalled 5.09 billion rubles – 40% less than a year earlier. Fifty deals were completed, half the number in the first half of 2025, with an average cheque of 113.2 million rubles. The largest share of investment went to artificial intelligence and machine learning – the sectoral focus mirrors the global one, but the scale does not.

Industry analysts compare current indicators to levels seen in 2009–2011. The logic of financing has changed structurally: with a high key interest rate, deposits and debt markets compete with venture returns, so investors require proven revenue, positive unit economics, and a clear path to profitability from startups, rather than a “promising idea.” The main sources of capital remain corporate venture, sector-specific funds, and club syndicates.

Conclusions for venture investors and funds

The agenda as of 29 July 2026 can be summarised in four points:

Record ≠ broad market. The aggregate $510 billion masks a narrowing funnel: capital is available to category leaders, not the average startup.
Concentration is a risk in itself. Portfolios whose returns depend on a few AI leaders need stress-testing against a scenario where one of them has a disappointing debut.
The exit window is open, but selective. Companies with valuations of $2–5 billion, sustainable revenue, and near-profitability have a genuine opportunity to use the current IPO cycle.
Infrastructure bets beat application-layer bets. Compute, energy, security, and physical AI offer a more defensive position than applications built on top of others’ models.

The market has entered a phase where an abundance of capital coexists with a scarcity of access to it. For venture funds and institutional investors, this means a return to fundamental discipline: quality of selection, valuation discipline, and sober liquidity planning – regardless of how impressive the headline numbers for the half-year may appear.