Startup and Venture Capital News — Tuesday, July 28, 2026: Anthropic IPO, Record $510 Billion, and Shift to Physical AI

Startup and Venture Capital News for July 28, 2026: Record Semester for Global Venture Market, Anthropic Listing Preparations, Capital Concentration in Megafunds, Investment Shift to Physical AI and Inference Infrastructure, Exit Recovery and Russian Market Contraction.

The venture market enters the last week of July 2026 in a state that is difficult to capture in a single word. By formal metrics, this is the best period in the industry's history: global venture investments in the first half of the year reached a record $510 billion according to Crunchbase, while the US market, per PitchBook and NVCA calculations, reached $412.7 billion — more than any full year prior. By internal structure, it is an extremely narrow market where money concentrates in a few dozen companies and about fifteen funds, while the average founder feels not a boom but a filter.

Key developments as of Tuesday morning, July 28, 2026

• Record semester. $510 billion in global venture investments in H1 2026 — a new high for any six-month period in industry history.
• AI dominance. In Q2, AI startups accounted for more than 70% of global volume, up from less than 50% a year earlier; in the US, artificial intelligence’s share reached $355.9 billion of $412.7 billion.
• Capital concentration. Over 81% of US venture dollars went into rounds of $100 million or more.
• LP market contraction. 16 megafunds raised nearly 70% of the $72.4 billion in new venture fundraising for the half-year.
• Liquidity return. 32 portfolio companies conducted IPOs with valuations above $1 billion; another 24 were acquired at more than $1 billion for a total of approximately $113 billion — a record quarter for M&A.
• Focus shift. Capital is moving from application software into hardware: inference chips, robotics, sensors, data for physical AI, and data-centre energy infrastructure.

Macro picture: record volume, deficit in breadth

The key paradox of the current cycle is that record-breaking startup and venture capital news describes an ever-smaller number of companies. In Q1 2026, global investment volume was roughly $297 billion, while four deals — OpenAI’s $122 billion round at $852 billion valuation, $30 billion for Anthropic, $20 billion for xAI, and $16 billion for Waymo — accounted for more than 63% of the quarterly total. In Q2, the distribution only slightly broadened.

For an investor, this signals a shift in the working hypothesis. Aggregate venture market figures are no longer an indicator of capital availability for the average portfolio company. The median Series B round in non-infrastructure segments closes more slowly than in 2021, with tighter covenants and more conservative liquidation preference structures. Early stage, however, has held up: in North America, seed and early-stage financing volume in Q2 exceeded $31 billion — nearly double year-over-year — but the number of deals was the lowest in five quarters. Cheques grew larger; the number of recipients shrank.

Fundraising: a two-speed market

A similar asymmetry appears on the LP side. Of the $72.4 billion raised by venture funds in the first half, about 70% went to 16 megafunds. Institutional investors continue to feel denominator pressure and have not fully restored liquidity after the 2022–2024 cycle, so they prefer to allocate capital to established franchises rather than expand the number of managers.

Practical implications for the venture investment market:

1. First- and second-time funds face increasing fundraising duration and lower target sizes.
2. Emerging managers increasingly shift to a model of SPVs and pledge funds instead of classic blind pool structures.
3. Megafunds gain the ability to lead rounds alone, reducing the role of syndicates and changing the bargaining position on valuation.
4. The secondary market becomes the main channel of interim liquidity for LPs until exits materialize.

Liquidity window: Anthropic IPO and the SpaceX effect

The week's main topic for managers is the public market. After SpaceX’s June listing on Nasdaq, where shares closed above the target on debut and then retreated below the offering price, the industry learned an important lesson: the window is open, but the scarcity premium fades quickly.

Attention now centres on Anthropic. The company filed a confidential S-1 in early June at a valuation of around $965 billion after its Series H round, and according to business media, held meetings with institutional investors in July; a listing is being discussed for October. Neither the price range nor the offering size has been officially confirmed. OpenAI, meanwhile, has pushed its expected public offering timeline to 2027, while Databricks has publicly ruled out a 2026 listing, discussing a private round at a $165–175 billion valuation.

For venture funds, the return of exits matters more than any record in fundraising. A record M&A quarter and three dozen IPOs with valuations above $1 billion, for the first time since 2021, create conditions where LP distributions begin to catch up with capital calls. This cycle, not absolute investment volumes, will determine fundraising in 2027.

Physical AI and the inference economy: where the front has shifted

The most notable structural shift in recent weeks is capital moving “down the stack.” Investors are funding not model-based applications, but what makes AI cheaper to operate and more applicable in the physical world.

A telling example is Etched’s round: $300 million Series C at a $10.3 billion valuation for specialized inference hardware. The logic is straightforward: model training created the first investment cycle, but recurring demand, token cost, energy consumption, and latency are determined at the execution stage. Meanwhile, European company Humanoid raised $152 million Series A at a $1.35 billion valuation for industrial humanoid robots with participation from Schaeffler and Bosch — strategic industrial investors are returning to venture as co-investors, not only as acquirers.

Physical AI has ceased to be a single category and has split into independent capital segments: inference computing and economics, hardware deployment and robotics, multimodal data from real interactions, and machine sensing. A separate direction is data-centre energy infrastructure — a segment where strategic minority investments are measured in billions of dollars.

Deals shaping the current feed

The common denominator of these deals is not sector fashion but the presence of a bottleneck. Financing goes to companies that reduce AI operating costs, improve its reliability in production, or embed it into industries with large recurring budgets: cybersecurity, insurance, healthcare, manufacturing.

Geography of capital: US, Europe, Asia, Middle East

The US maintains dominance: about 88% of global AI capital goes to American companies. North America attracted $392 billion in the half-year. Nevertheless, geography is diversifying by verticals rather than by volume.

• Europe is reclaiming positions in defence technologies, fusion energy, and industrial robotics, relying on strategic capital from industrial corporations and state development institutions.
• Asia is strengthening in data infrastructure for robotics; Singapore is consolidating as a deep-tech hub with global ambitions.
• The Middle East has shifted from an LP role to a lead-investor role: sovereign entities and corporate venture arms are leading rounds in AI infrastructure directly.

Russia and CIS: market contracts faster than the global cycle

Russian dynamics move against the global trend. According to the Venture Guide platform, in the first half of 2026, investment volume in domestic startups amounted to approximately 5.2 billion rubles — 39% less year-over-year, with the number of disclosed deals roughly halved to 52. Moscow concentrates about 64% of investments and 63% of deals, while the regional ecosystem is effectively stagnating. The largest volume goes to projects in artificial intelligence and machine learning. The consensus forecast for the year is 11–13 billion rubles with 110–130 disclosed deals, i.e., a level comparable to the anti-record year of 2023.

A key feature of the local market is a shift toward full buyout deals and corporate demand for ready-made integrable solutions, instead of the classic venture cycle of successive rounds followed by a public market exit.

Risks and agenda for managers

1. Concentration risk. When 63–72% of quarterly volume is formed by a handful of deals, industry indices stop reflecting the state of the median portfolio.
2. Infrastructure overvaluation risk. Early contractual commitments in semiconductors are easier to obtain than to confirm with manufacturing reliability.
3. Post-listing dynamics risk. The pullback in share prices after the largest offerings affects the mark-to-market of the entire late stage.
4. LP liquidity risk. Until distributions fully recover, fundraising for new funds will remain two-speed.
5. Commoditization risk. At the application layer, without proprietary data, distribution, or switching costs, defensibility erodes faster than revenues grow.

Frequently asked questions

What is the volume of the global venture market in 2026? In the first half of 2026, global venture investments reached a record $510 billion, exceeding any previous six-month result.

What share does artificial intelligence account for? In the second quarter, AI startups accounted for more than 70% of global volume; in the US, about 86% of all venture dollars in the half-year.

Is the IPO window open for technology companies? Yes, but selectively: more than 30 venture-backed companies went public with valuations above $1 billion, yet post-listing dynamics of the largest deals show a rapid compression of the premium.

Where is the focus of venture investors shifting? To inference infrastructure, semiconductors, robotics, sensors, data for physical AI, and data-centre energy infrastructure.

Conclusion

The market on July 28, 2026 is record volumes with record selectivity. Capital is available, but it buys not an idea or growth rate, but control over a bottleneck: computing economics, proprietary data, physical deployment, or a regulated process with a large budget. For venture funds, the coming months will be determined not by the next mega-round, but by the industry’s ability to convert record investments into real distributions — through the Anthropic IPO, a record M&A cycle, and the secondary market. It is at this intersection that the premium is being formed in 2026.