March 2026's Largest Startup Raises — Where Institutional Capital Is Flowing

TL;DR: March 2026 is one of the most active months for startup capital formation in venture history. Q1 2026 is on pace to surpass the entirety of 2024's venture capital activity as allocators re-risk into growth, and AI continues to dominate the largest private rounds on record. OpenAI added another ~$10 billion to its record-setting raise, taking the total past $120 billion. Nscale closed Europe's largest equity round at $2 billion. Shield AI secured $2 billion in combined equity and preferred financing for defense autonomy. AMI Labs — co-founded by Yann LeCun — launched with Europe's largest seed round ever at $1.03 billion. And robotics companies collectively absorbed more than $1.2 billion in a single week. For GPs raising new vehicles and family offices calibrating deployment plans, the signal is clear: private capital is concentrating at the top of the AI stack, but enabling infrastructure, physical AI, and defense tech are opening parallel allocation channels that didn't exist two years ago.

The Mega-Rounds: Where did $500M+ checks go in March 2026?

1) OpenAI adds ~$10B to its record-setting raise — total now exceeds $120 billion

OpenAI is raising an additional approximately $10 billion in commitments on top of its previously announced $110 billion round, taking the current fundraise past $120 billion and cementing one of the largest private financings ever completed. Amazon, Nvidia, and SoftBank anchor the investor group. The round — which was first disclosed in late February — represents a structural shift in how late-stage AI companies raise: not as a single close, but as an ongoing capital formation event with rolling commitments from strategic and sovereign allocators.

Why it matters for LPs. OpenAI is functioning less like a startup raise and more like a sovereign wealth fund-grade infrastructure allocation. For limited partners evaluating their managers' exposure to foundational AI, the question is no longer whether they have OpenAI in portfolio — it's whether they have anything besides OpenAI. February 2026 saw a record $189 billion in global VC deployed, yet roughly 83% of that went to just three companies: OpenAI, Anthropic, and Waymo. That is a concentration-risk signal that every allocator and investment committee should be stress-testing.

2) Nscale closes $2B Series C at $14.6B — Europe's largest equity round

UK-based hyperscale AI infrastructure provider Nscale closed a $2 billion Series C at a $14.6 billion valuation — described by the company and independent coverage as the largest Series C and one of the largest equity rounds ever completed in Europe. The valuation represents a 4x+ increase from the $3.1 billion mark set at its September 2025 Series B. Nvidia, Citadel, Dell, Nokia, Jane Street, Aker ASA, and 8090 Industries participated, with Goldman Sachs and J.P. Morgan acting as placement agents.

Why it matters for LPs. The placement-agent involvement from Goldman and JPM is the structural tell here. GPU data center capacity is now being underwritten with the same balance-sheet capital normally reserved for energy, telecom, and transport infrastructure. For family offices and pensions that historically allocated to infrastructure through dedicated vehicles, AI compute is increasingly competing for the same allocation bucket — not the venture bucket. Fund managers positioning around AI infrastructure should understand that they're not just competing with other VCs for LP attention; they're competing with infrastructure GPs, credit funds, and direct co-investment desks.

3) Shield AI secures $2B in combined financing at $12.7B — defense AI reaches blue-chip PE and bank capital

San Diego-based defense autonomy platform Shield AI announced a $1.5 billion Series G at a $12.7 billion post-money valuation, alongside an additional $500 million of fixed-return preferred equity backed by Blackstone, for a total $2 billion financing package. Advent International led the equity round, with JPMorgan Chase's Strategic Investment Group as co-lead. Part of the proceeds will fund Shield AI's planned acquisition of Aechelon Technology, a defense simulation specialist. The round more than doubles Shield AI's valuation from a year ago.

Why it matters for LPs. This is the deal that formally moves defense AI out of specialist-only territory. When Advent International — a global PE firm managing $100B+ — co-leads alongside JPMorgan's strategic investment arm, it signals that dual-use autonomy and mission-critical software are now within the core mandate of blue-chip private equity and bank-aligned capital. For GPs raising defense-adjacent vehicles, the Shield AI syndicate is proof of concept that institutional LPs are ready to underwrite the category at scale. For family offices that have been watching defense tech from the sidelines, the Blackstone preferred tranche also shows that fixed-return instruments can provide defense exposure without pure equity risk.

4) AMI Labs launches with $1.03B seed — Europe's largest seed round in history

AMI Labs (Advanced Machine Intelligence), co-founded by Turing Award winner and former Meta AI chief Yann LeCun, emerged with a $1.03 billion seed round valuing the Paris-based company at roughly $3.5 billion. AMI is developing "world models" — AI designed to learn structured representations of the physical world rather than relying on text-only large language models. Bezos Expeditions, Nvidia, Samsung, and Temasek participated, alongside sovereign and institutional backers.

Why it matters for LPs. AMI represents the most well-funded bet against the autoregressive text-prediction paradigm that powers ChatGPT, Claude, and Gemini. For allocators with concentrated LLM exposure, AMI is a hedging signal — a data point suggesting that the next generation of AI may look structurally different from the current one. The Temasek and sovereign participation also confirms that state-linked capital is willing to underwrite fundamental AI research bets at seed stage, not just late-stage infrastructure. GPs building theses around embodied AI, robotics, or computer vision should track AMI's technical direction closely — its JEPA architecture could reshape which downstream application companies become investable.

5) Sierra Space closes $550M at $8B — space infrastructure draws family and sovereign capital

Colorado-based Sierra Space, which designs and manufactures satellites, spacecraft, and space subsystems, secured $550 million in equity financing led by LuminArx Capital Management. General Atlantic, Coatue, Moore Strategic Ventures, and Andalusian Private Capital joined the round. The five-year-old company's valuation now sits at $8 billion.

Why it matters for LPs. Space remains one of the few capital-intensive frontier categories where family offices, sovereign funds, and growth equity regularly co-invest in the same rounds. For fund-of-funds and OCIOs building thematic exposure, Sierra Space is a representative example of how defense-adjacent infrastructure can sit in a portfolio alongside pure defense holdings like Shield AI — providing exposure to dual-use technology without pure military dependency.

6) Legora raises $550M Series D at ~$5.55B — legal AI reaches late-stage escape velocity

Swedish legal AI startup Legora raised $550 million in a Series D led by Accel, tripling its valuation to approximately $5.55 billion. Benchmark, Bessemer Venture Partners, General Catalyst, Iconiq, Redpoint Ventures, and Y Combinator all participated.

Why it matters for LPs. Legora is the clearest proof point that vertical AI SaaS — here in legal services — can attract late-stage capital at software-style multiples while being positioned as workflow infrastructure rather than a model company. For LPs evaluating whether their managers are overly indexed to foundational model companies, Legora represents the application-layer alternative: enterprise-grade AI revenue without the compute capex burden. Goldman Sachs Growth Equity's consecutive weeks of backing regulated professional-services AI (Grow Therapy, Fieldguide, Oro Labs, and now the Legora syndicate overlap) confirms this as an institutional-grade category, not a venture experiment.

7) Quince secures $500M Series E at $10.1B — non-pure-play AI can still raise mega-rounds

E-commerce and supply-chain automation company Quince closed a $500 million Series E led by Iconiq Capital at a $10.1 billion valuation. Quince operates a vertically integrated "manufacturer-to-consumer" model with AI-driven logistics and merchandising capabilities.

Why it matters for LPs. Quince is a useful benchmark for allocators asking whether every mega-round in 2026 requires an "AI-first" thesis. It doesn't. Quince is raising on unit economics, margin structure, and operational leverage — with AI as an enablement layer, not the core product. For LPs tired of pure-model exposure, Quince-style deals offer a reminder that durable DPI can come from companies that use AI without being defined by it.

8) Nexthop AI raises $500M Series B at ~$4.2B — AI networking becomes its own investable category

AI networking infrastructure startup Nexthop AI raised $500 million in a growth round led by Lightspeed Venture Partners, with participation from Andreessen Horowitz, Altimeter Capital, and Kleiner Perkins. The company builds specialized switching technology and software for AI data centers, targeting latency and power consumption reduction.

Why it matters for LPs. Nexthop is a pick-and-shovel play on AI infrastructure that sits below the model layer and above raw compute. For allocators mapping the AI capital stack in detail — which an increasing number of endowments and family offices are now doing — networking is a distinct layer that was essentially un-investable two years ago and now commands $500M rounds. Lightspeed and a16z co-investing signals consensus that this is a category, not a single-deal anomaly.

9) Mind Robotics closes $500M Series A — Rivian spinout raises at ~$2B on physical AI thesis

Rivian spin-out Mind Robotics closed a $500 million Series A co-led by Accel and Andreessen Horowitz, following a $115 million seed in late 2025. Total capital raised now sits at roughly $615 million, with the company valued near $2 billion. Mind Robotics develops AI-enabled industrial robotics for manufacturing environments.

Why it matters for LPs. A $500 million Series A signals that physical AI has decisively left the R&D phase and entered scale-up territory. For family offices and institutional LPs who have historically avoided hardware-centric deep tech due to capital intensity and long payback horizons, the Mind Robotics syndicate (Accel + a16z at co-lead) sends a specific message: the valuation step-ups, check sizes, and syndicate quality in robotics now look more like software infrastructure than hardware ventures.

10) Rhoda AI emerges from stealth with $450M Series A at ~$1.7B

Industrial robotics intelligence platform Rhoda AI exited stealth with a $450 million Series A led by Premji Invest, with Temasek and other global institutions among its backers. Rhoda's FutureVision platform trains robots from large-scale video data to handle unstructured industrial environments.

Why it matters for LPs. Rhoda and Mind Robotics together account for nearly $1 billion in Series A capital deployed into industrial robotics in the same month. Combined with Skild AI's $1.4B Series C in January, this creates a clear pattern: robotics companies collectively raised more than $1.2 billion in a single week in mid-March. For allocators building a physical AI allocation view, the investable universe went from sparse to crowded in two quarters.

11) Replit raises $400M at $9B — AI-assisted coding reaches infrastructure-tier valuation

Developer platform Replit raised $400 million in new funding, tripling its valuation to approximately $9 billion. The company is approaching $1 billion in projected annual recurring revenue as it doubles down on AI-assisted "vibe coding" tools.

Why it matters for LPs. Replit's $9 billion valuation — a 3x step-up in six months — confirms that AI application companies can achieve infrastructure-tier valuations when they demonstrate enterprise-scale adoption and a clear path to durable cash flows. For late-stage growth investors, Replit represents the kind of revenue-backed AI bet that's increasingly preferred over pre-revenue model companies with higher compute exposure. The approaching $1B ARR milestone is a valuation floor that appeals to crossover funds and growth-stage family office direct investment programs.

Notable $100M–$400M rounds worth tracking

Beyond the mega-rounds, several mid-stage financings offer additional signal for LPs tracking institutional capital flows and venture capital trends in Q1 2026.

Cloaked — $375M Series B. Privacy and identity platform, led by General Catalyst and Liberty City Ventures. Consumer data protection at growth-equity scale — a category that didn't exist at this check size three years ago.

Cambridge Mobile Telematics — $350M. Telematics and mobility AI for road safety and risk scoring, led by The Rise Fund and Allianz X, with State Farm participating. The insurance-adjacent LP base here (Allianz, State Farm) signals that insurance balance sheets are entering venture-stage mobility deals through strategic channels — a pattern family offices co-investing alongside insurers should note.

Reflection — targeting $2.5B at $25B. Open-source AI models backed by Nvidia. Still in process at time of publication but, if completed, would be one of the largest raises for an open-weight model company — proof that proprietary isn't the only investable AI paradigm.

eMed — $200M Series A at $2B+. Miami-based AI-driven telehealth platform scaling a GLP-1 weight-loss program in partnership with CVS Caremark and Aon. Healthcare AI with payer-channel distribution — a structure that appeals to family offices seeking revenue-backed health exposure without pure biotech risk.

Cents — $140M Series C. Laundry-operations software led by Sumeru Equity Partners. One of the largest single software investments in the laundry vertical. A pure vertical SaaS play with $1 billion in annual processed payments — the kind of "boring" infrastructure deal that often outperforms on DPI.

Granola — $125M Series C at $1.5B. AI productivity platform led by Index Ventures, transitioning from transcription toward an enterprise workflow layer. Continues to feature in allocator conversations around applied productivity AI.

Kai — $125M. Agentic AI cybersecurity platform led by Evolution Equity Partners. Built for machine-speed incident response. Responds to the CISOs' reality that AI-enabled attack surfaces require AI-enabled defenses.

Dash0 — $110M Series B at ~$1B. Observability and monitoring, led by Balderton Capital. Extending the AI observability narrative to European software infrastructure — and crossing the $1B valuation mark.

Gimlet Labs — $80M Series A. AI inference-infrastructure optimization, led by Menlo Ventures. Another building block in the AI infrastructure stack that LPs are increasingly mapping layer by layer.

What does March's capital formation tell LPs and allocators?

AI infrastructure is becoming a sovereign-level asset class

Nscale's $2 billion Series C, Nebius's multi-billion-dollar convertible financing for AI data centers, and OpenAI's $120-plus billion cumulative round all share a common pattern: sovereign wealth funds, strategic corporates, and large alternative managers are treating GPU compute, energy access, and data center real estate as long-duration assets akin to pipelines or telecom networks. For LPs focused on institutional capital flows, the implication is that "AI infrastructure" is no longer just an IT budget line — it's an investable category spanning credit, growth equity, and infrastructure funds. This requires deliberate allocation decisions at the portfolio-construction level rather than opportunistic single-deal exposure.

Robotics entered its mega-round era

Mind Robotics ($500M Series A), Rhoda AI ($450M Series A), Sunday ($165M at unicorn valuation for household robots), and Oxa ($103M for autonomous logistics) collectively represent more than $1.2 billion of capital deployed into robotics in a single week. Combined with Skild AI's $1.4B Series C in January, 2026 is on pace for $20 billion or more in robotics funding. For LPs and family offices evaluating their venture exposure, the message is that physical AI — robots operating in warehouses, factories, and logistics environments — has moved from R&D-stage deep tech to scale-up platforms with check sizes, valuation step-ups, and syndicate quality that look more like software infrastructure deals.

Defense tech reached mainstream institutional capital

Shield AI's $1.5 billion Series G — paired with $500 million of preferred equity and co-led by Advent International and JPMorgan's strategic investment arm — demonstrates that global PE sponsors and bank-associated platforms are now comfortable underwriting dual-use autonomy and mission-critical software. Coupled with Sierra Space's $550 million round, defense-adjacent AI is likely to appear with increasing frequency in generalist growth and crossover portfolios rather than remaining confined to specialist defense vehicles. For family offices that have historically relied on public-market defense primes (Lockheed, Raytheon, Northrop) for exposure, the private-market alternative is now liquid enough to warrant direct engagement.

Concentration risk is the allocator's quiet problem

The headline numbers for global venture activity mask an extreme concentration of capital at the top of the stack. February 2026 saw a record $189 billion in global VC deployed, yet around 83% went to just three companies: OpenAI, Anthropic, and Waymo. Combined, those three absorbed roughly $156 billion. For LPs, this raises a clear concentration-risk question: are their managers structurally exposed only to the very top of the AI layer through secondary positions and late-stage growth allocations, or do they also have ownership in the enabling layers — networking (Nexthop AI, Gimlet Labs), infrastructure (Nscale, Nebius), applied vertical AI (Legora, Granola, Kai), and physical AI (Mind Robotics, Rhoda AI) — that may offer more diversified return profiles across the 2026 vintage year?

What's coming: Altss expands into company and startup intelligence

In the coming months, Altss is expanding beyond LP and family office intelligence into comprehensive company and startup data — bringing 10 million+ company profiles onto the platform. This means the same OSINT-driven methodology that powers verified allocator profiles will apply to the companies those allocators are investing in.

For GPs, this closes a loop that has historically required multiple platforms: you'll be able to identify which LPs are allocating to AI infrastructure, see the companies they've backed, map competitive rounds, and track portfolio overlap — all within Altss. For family offices running direct investment programs, it means startup discovery, LP peer benchmarking, and co-investment tracking in a single intelligence layer.

The expansion covers private companies across all stages (seed through pre-IPO), industries, and geographies. Company profiles will include funding history, investor syndicates, team data, sector classification, and OSINT-derived signals — built to the same verification standard as Altss's existing LP coverage. It represents the largest single product expansion since Altss added institutional LP coverage in February 2026 and transforms the platform from an LP intelligence tool into a full private-markets intelligence layer.

More details will be published as the rollout begins. If you want early access, book a demo.

Altss lens: where the warm paths actually sit

For fundraising teams reading this as a deal-sourcing signal rather than a market recap, here's where we see the actionable paths:

If you're raising around AI infrastructure. The Nscale and Nexthop syndicates — Nvidia, Citadel, Dell, Jane Street, Lightspeed, a16z — represent allocators who are now repeat-deploying into AI compute and networking. Many of these investors also back adjacent infrastructure plays at earlier stages. Altss tracks which of these institutional investors and family offices have active mandates in AI infrastructure and at what check size, so you can identify who's genuinely in-market versus who's already fully deployed for the vintage.

If you're raising around robotics or physical AI. Bezos Expeditions, Temasek, Premji Invest, Accel, a16z, and SoftBank now form a core syndicate group for physical AI. The overlap between the Skild AI (January), Mind Robotics, and Rhoda AI (March) rounds gives fundraising teams a direct map of who wrote nine-figure checks and who's likely to evaluate the next round. Altss surfaces which family offices participated in these rounds and whether they have co-investment capacity remaining.

If you're raising around defense or dual-use. Advent International, JPMorgan's Strategic Investment Group, Blackstone (via the Shield AI preferred tranche), and the existing specialist defense VC syndicates represent a blended LP base that didn't exist in defense two years ago. The entry of blue-chip PE and bank-aligned capital means defense GPs can now approach institutional LPs who previously viewed defense as off-mandate.

If you're raising around vertical AI SaaS. Legora's syndicate — Accel, Benchmark, Bessemer, General Catalyst, Iconiq, Redpoint, YC — is a who's-who of growth-stage venture. The Goldman Sachs Growth Equity pattern (consecutive deployments into regulated vertical AI) confirms that institutional growth capital is actively seeking applied AI with enterprise revenue, not just foundational model exposure.

Methodology

This article uses the same OSINT methodology that underpins all Altss intelligence products, documented in full in the OSINT Intelligence Framework. Deal data was sourced from verified public disclosures including company announcements, SEC filings, regulatory records, and coverage from Crunchbase, TechCrunch, Reuters, the Financial Times, and sector-specific sources. All figures — amounts raised, valuations, and investor names — were cross-referenced against at least two independent sources before publication. Where a figure could not be independently verified or was described as "reportedly" in source material, that qualifier is preserved.

This article does not constitute investment advice. It is a market intelligence summary for informational purposes.

Known gaps and caveats. Several March 2026 rounds were still in process at time of publication (Reflection's $2.5B raise, OpenAI's rolling commitments). Final amounts may differ from figures reported here. Some rounds did not disclose whether capital was structured as equity, convertible debt, or structured preferred — which matters for valuation comparisons. Where structure is unknown, we note it. Figures for robotics "weekly" capital deployment are aggregated from Crunchbase weekly roundups for the week of March 7–13 and may not capture all global robotics deals in that period.

FAQ

How much total capital was deployed in March 2026 mega-rounds? The eleven $400M+ rounds covered in this article represent more than $9 billion in disclosed capital. Including the $100M–$400M rounds, total coverage exceeds $11 billion. This does not include OpenAI's rolling commitments, which would add another ~$10 billion.

Which investors appeared most frequently across March 2026 deals? Andreessen Horowitz (Nexthop AI, Mind Robotics), Accel (Legora, Mind Robotics), Nvidia (OpenAI, Nscale, AMI Labs, Reflection), and Iconiq (Legora, Quince) each appeared in multiple mega-rounds. Temasek participated in both AMI Labs and Rhoda AI, reinforcing its position as one of the most active sovereign allocators in AI and robotics.

Is March 2026 actually a record month for startup funding? Q1 2026 is on pace to surpass all of 2024's VC activity, driven primarily by the massive February rounds (OpenAI $110B, Anthropic $30B, Waymo $16B). March's mega-rounds are smaller individually but more widely distributed across sectors — AI infrastructure, defense, robotics, legal AI, e-commerce — suggesting a broadening of capital deployment beyond pure foundational model companies.

How does Altss track these rounds? Altss uses OSINT — open source intelligence — to monitor capital formation events in real time: company announcements, regulatory filings, investor disclosures, and news signals across 40+ jurisdictions. For more detail on the methodology, see OSINT Methodology for LP Intelligence.

Where can fund managers find LP data for the investors in these rounds? Altss tracks 9,000+ family offices and 30,000+ institutional LPs worldwide — including many of the allocators referenced in this article. The platform maps mandates, check sizes, geography, sector focus, and recent allocations. For a detailed comparison of how Altss compares to other LP data providers, see The Top 5 Family Office & Fundraising Intelligence Platforms. Book a demo to see how it works.

What is the company intelligence expansion mentioned in this article? Altss is expanding into company and startup data, bringing 10 million+ company profiles onto the platform. This means GPs and family offices will be able to track LP intelligence and company intelligence — funding rounds, investor syndicates, competitive mapping — in a single platform. The rollout begins in the coming months. Request early access.