AI fund formation baseline: where new capital is crowding the agent and infra stack (2026)

Launch baseline: a single-week sample did not produce enough verified fund-formation signals to show the channel's intended brief format, so this first issue widens the lookback to map active AI fund capital. Future briefs will tighten back to the weekly window and treat older funds as context only.

AI capital formation is no longer a simple question of "who has an AI thesis." The more useful question is which pools of capital can win a founder's attention: specialist seed investors with AI-native judgment, corporate-backed platforms with distribution, or growth funds that can finance compute-heavy scale-up.

The market backdrop supports that split. OECD's 2026 policy brief estimates that AI companies took 61% of global VC investment in 2025, or $258.7 billion of $427.1 billion in total VC investment, while IT infrastructure and hosting attracted $109.3 billion of AI VC in 2025. 1 Ropes & Gray's H1 2025 AI report adds the allocator tension: overall private-capital fundraising for AI was down year over year, but VC funds still accounted for most AI-focused private capital, and AI-focused VC represented 31% of all VC capital raised in H1 2025 when PE funds are excluded. 2

Fund formation signals

Fund / manager	Size and status	Strategy, stage, and geography	LP signal	Granola-adjacent read
Cathay Innovation Fund III	$1 billion final close, announced May 27, 2025. 3	Multistage vertical AI fund investing globally across Europe, the U.S., and Asia, with stated focus on healthcare, fintech, consumer, and mobility / energy. The fund says it can invest up to $100 million in one startup. 3	Backed by more than 20 corporations and institutional investors; named corporate backers include Sanofi, TotalEnergies, Valeo, BNP Paribas Cardif, Groupe SEB, Groupe ADP, Vale Ventures, and Copec WIND Ventures. 3	More likely to compete at later Series A and growth stages than pre-seed. For AI agents and workplace tools, the threat is corporate distribution: Cathay can pair vertical AI companies with sector buyers before a purely financial investor can.
AI Fund II	Approximately $190 million oversubscribed Fund II, announced May 5, 2025. 4	Venture studio model for 0-to-1 AI startups. AI Fund says it builds companies around the application and software infrastructure layers of the AI stack, including agentic AI, reasoning models, and multimodal AI. 4	Corporate LPs include AES, HP, Mitsui, Mitsubishi, QBE, and TELUS Global Ventures, alongside Sequoia Capital and NEA. 4	Directly relevant to agent workflow startups. The studio can create companies with built-in customer discovery, proprietary data access, and early distribution through corporate LPs, which can be more valuable than a larger first check.
Conviction Partners Fund II	$230 million second fund, reported January 31, 2025. 5	Early-stage AI investor focused on AI companies; Dakota describes the mandate as AI, enterprise software, and generative AI. 6	TechCrunch reported that the fund is more than double Conviction's $101 million debut fund and that Mike Vernal joined as GP after prior roles at Sequoia and Facebook. 5	High-likelihood competitor or co-investor for seed and Series A AI apps. Conviction's brand around Mistral, Harvey, and Sierra-style companies makes it especially relevant for AI productivity, legal, sales, and knowledge-work agents.
8VC Fund VI	$998 million in new LP capital, announced March 4, 2025. 7	Broad frontier-tech fund across healthcare, defense, logistics, enterprise, education, and life sciences. 8VC explicitly names AI solutions in healthcare and AI blurring the line between technology and services in enterprise. 7	Large flagship LP commitment, not a narrow specialist close. The signal is scale: 8VC can support company formation, regulated-industry access, and follow-on capital across several AI-heavy verticals. 7	Relevant where AI agents touch regulated workflows: healthcare operations, logistics, government, education, legal, and financial services. It may not lead every AI-app seed, but it can become a strong partner when workflow products need industry access.
AIX Ventures Fund II	$202 million second fund, announced February 13, 2024. 8	Early-stage AI fund focused on AI applications and infrastructure that shape the future of work. 8	AIX says its practitioner network and portfolio include Hugging Face, You.com, Weights & Biases, Perplexity, and Chroma. 8	Still relevant dry powder for agent and infrastructure founders. AIX overlaps with developer tools, retrieval, AI search, and workplace automation, which puts it close to many Granola-adjacent seed rounds.
Radical Ventures Growth Fund I	Nearly $800 million growth fund announced August 19, 2024. 9	Growth-stage AI fund for AI-first products and AI-enhanced businesses. 9	The signal is specialist scale at growth stage: Radical already had early-stage AI relationships and positioned the vehicle as a way to finance the next phase of scaling. 9	Less relevant to first institutional checks, but important for follow-on planning. A seed investor in AI agents should know when Radical could become a Series B or growth-stage partner, especially for infra-heavy companies.

What the capital is telling founders

Seed AI apps are still crowded, but check writers are specializing. Conviction, AIX, and AI Fund all sit close to the same terrain: AI-native applications, future-of-work products, developer tools, and company formation around new model capabilities. Their edge is not just capital size. It is whether they can help a founder pick a wedge before the category becomes obvious.

Corporate LPs are turning into distribution assets. Cathay and AI Fund both emphasize corporate relationships. For AI agents, this matters because enterprise adoption depends on workflow access, data permissions, procurement patience, and credibility with buyers. A corporate-backed fund can become a source of pilots or design partners, but it can also bias a startup toward specific vertical problems.

Infrastructure gravity is changing fund positioning. OECD's 2025 data shows infrastructure and hosting as the largest AI VC category by dollars, with $109.3 billion invested in 2025. 1 That pushes even app-focused investors to ask harder questions about inference cost, data pipelines, model routing, reliability, and distribution. A fund that cannot underwrite infrastructure dependencies may struggle to win the best agent companies.

Growth capital is increasingly part of the seed pitch. Radical's growth fund and 8VC's nearly $1 billion flagship show why seed investors need a later-stage map. If an AI app works, the next constraint may be sales capacity, compliance, compute commitments, or a platform acquisition threat. Founders will ask whether early investors can introduce follow-on partners with relevant appetite.

Implications for AI dealmaking

For a Granola-adjacent investor, the most competitive part of the market is likely not frontier models. It is the layer where AI agents turn messy work into repeatable workflows: meetings, research, legal drafting, customer operations, sales follow-up, software development, and internal knowledge management.

Three fund types are most likely to appear across the table:

AI-native seed specialists such as Conviction and AIX, which can credibly evaluate product taste, model leverage, and founder-market fit in AI apps.
Studio and operator-led builders such as AI Fund, which may originate companies before the open market sees them.
Corporate-network and vertical platforms such as Cathay and 8VC, which can win when the company needs customers, data, or regulated-market access.

The practical takeaway: founder access will depend less on saying "we invest in AI" and more on proving a specific advantage. For AI agents, that advantage could be a proprietary distribution channel, a sharper view on workflow data, a known follow-on path, or the ability to help founders avoid selling a wrapper when they need a system of record.

Watch list for the next weekly brief

Next issues should prioritize fresh fund-formation evidence over broad AI market data:

New SEC Form D filings for AI specialist funds, operator-led funds, and emerging managers.
Fund announcements that disclose size, stage, sector, geography, and LP composition.
Partner posts that reveal strategy shifts toward agents, vertical AI, inference infrastructure, or workflow automation.
LP conference agendas and allocator commentary that name AI infrastructure, emerging managers, or specialist venture mandates.
Funds showing up near Granola-adjacent companies through co-investments, portfolio overlaps, or repeat founder referrals.

AI fund formation baseline: where new capital is crowding the agent and infra stack

Fund formation signals

What the capital is telling founders

Implications for AI dealmaking

Watch list for the next weekly brief

참고 출처

관련 콘텐츠