Back
Table of Contents
Six US Series B+ rounds closed between May 4 and May 11, 2026, totaling $672.0M. AI infrastructure and vertical AI dominated the week, with Corgi's $160M insurtech round, Panthalassa's $140M wave-powered compute play, and three Series B closes from Avoca, DeepInfra, and Parallel Web Systems together accounting for over $470M. The week's smallest deal, Moment Energy's $40M Series B, anchored the climate-tech end of the table.
The late-stage activity sits inside a broader US funding picture this week. Growth-stage rounds pulled in $1.82B across three deals, and Series A activity added another $162.8M across five companies. Series B+ rounds clustered around AI verticals (insurance, home services, inference, agentic search) and infrastructure bets that need real capital to ship physical product. Five of the six rounds were $100M or larger, which tells you something about where institutional capital is concentrating right now.
1. Corgi Raises $160M For AI-Native Commercial Insurance
Deal Overview
- Stage: Series B
- Sector: Fintech (insurtech)
- Geography: United States
- Round size: $160M
- Valuation: $1.3B
Investor Profile
TCV led the round, with participation from Oliver Jung, Leblon Capital, Kindred Ventures, Repeat VC, and Y Combinator. TCV brings deep fintech and growth-stage operating experience, including investments in companies like Toast, Klarna, and Revolut. The syndicate is a mix of growth capital and operator angels.
This round closed just four months after Corgi's Series A. That cadence is among the fastest Series A-to-unicorn progressions insurtech has seen.
Company and Leadership
Corgi was founded in 2024 by Emily Yuan and Nico Laqua. The company is Y Combinator-backed and received regulatory approval as a licensed insurance carrier in July 2025. Total funding now exceeds $268M.
Problem and Opportunity
US commercial property and casualty insurance runs on legacy stacks where underwriting, policy administration, and claims live on separate 20-year-old systems. Commercial trucking alone is a $50B premium market notorious for inefficient pricing, slow claims, and high loss ratios.
The first wave of insurtech (Lemonade, Root, Hippo) tried to disrupt personal lines and largely disappointed. Commercial lines, where data complexity and underwriting depth create real defensibility, remained mostly untouched.
Product and Technology
Corgi operates as a full-stack AI-native carrier, not a managing general agent. It owns the licensed carrier entity and runs underwriting, claims, policy admin, and pricing on one unified AI-driven platform. The first vertical is commercial trucking.
Owning the carrier means Corgi owns the loss data on its own book of business. That data feeds the underwriting and pricing models directly. Replacing three legacy systems with one platform creates unit economics that incumbents can't match without a ground-up rebuild.
Use of Proceeds and Vision
The capital funds expansion into additional commercial verticals beyond trucking, scaling the underwriting and claims AI, and hiring across product, engineering, and distribution. Strategic acquisitions of MGAs or legacy books are likely on the table to accelerate vertical entry.
Corgi's positioning is to become the carrier-of-record for commercial lines across multiple verticals, replacing the fragmented MGA plus reinsurance plus legacy carrier model.
Market Context
US commercial P&C is a $400B+ premium market dominated by Travelers, Chubb, and Liberty. The 2026 insurtech thesis has shifted to AI-native operators in commercial lines. Direct competitors include Hippo, Lemonade, Coalition in cyber, and Pie Insurance in workers' comp, though none operate as full-stack commercial carriers at Corgi's scope.
Startups like yours already closed their rounds with us.
Founders across every stage and industry. Here's what it took.
- Raised $7.6M for Swiipr Technologies
- Raised $0.5M for Ap Tack
- Raised €0.5M for Ivent Pro
2. Panthalassa Raises $140M For Ocean-Powered AI Inference
Deal Overview
- Stage: Series B
- Sector: Climate tech (AI infrastructure)
- Geography: United States (Pacific Northwest)
- Round size: $140M
- Lead: Peter Thiel
Investor Profile
Peter Thiel led the round. The syndicate reads as a who's-who of Silicon Valley: John Doerr, TIME Ventures (Marc Benioff), SciFi Ventures (Max Levchin), Susquehanna Sustainable Investments, Hanwha Asset Management, Fortescue Ventures, Future Positive, Super Micro Computer, Sozo Ventures, Dylan Field, Planetary VC, Leblon Capital, and Portland Seed Fund.
The combination of Thiel-Doerr-Benioff plus Korean strategic capital (Hanwha) and Super Micro as a hardware partner signals serious institutional belief in a thesis that is otherwise unproven.
Company and Leadership
Panthalassa is a Pacific Northwest climate-tech and ocean-energy company headquartered near Portland, Oregon. This Series B is one of the most high-profile climate-infrastructure rounds of 2026.
Problem and Opportunity
AI inference demand is the binding constraint of the 2026 to 2030 hyperscaler era. Terrestrial grid capacity, permitting timelines, and water availability are bottlenecks that don't bend easily. Most hyperscalers are signing power purchase agreements years in advance just to keep up.
Adjacent bets to solve this include nuclear SMRs (Oklo, X-energy), gas peakers, and stratospheric or space-based compute. Wave-powered offshore compute is the most ambitious of these.
Product and Technology
Panthalassa deploys floating offshore compute nodes (the Ocean-3 series) powered by ocean-wave energy. Power is generated by proprietary wave-energy conversion onboard each node and used directly to run AI accelerators. Inference tokens return to land via satellite uplink. Pilot deployment is planned in the northern Pacific in 2026, with commercial deployments targeted for 2027.
The stack is vertically integrated: wave-energy generation plus AI compute with no grid dependency, no land permitting, and no water-cooling constraints. Once deployed, marginal energy cost approaches zero.
Use of Proceeds and Vision
The round funds completion of the pilot manufacturing facility near Portland, accelerated deployment of Ocean-3 nodes, and validation of inference economics at sea. The bet is that AI inference demand will exceed terrestrial capacity within five years, forcing compute offshore.
Market Context
Pilot manufacturing near Portland also creates a US-domiciled, sovereign-aligned offshore compute supply chain, which matters for defense and intelligence customers. This is high-beta infrastructure investing. The Thiel-Doerr stack suggests the asymmetric payoff is what matters here, not near-term revenue.
3. Avoca Raises $125M For Trades-Industry Voice AI
Deal Overview
- Stage: Series B
- Sector: AI (vertical, home services)
- Geography: United States (New York / Santa Barbara)
- Round size: $125M
- Valuation: $1B
Investor Profile
General Catalyst and Meritech co-led, with Kleiner Perkins, Amplify Partners, and Y Combinator participating. Kleiner led the prior Series A. General Catalyst's involvement signals enterprise-scale ambition; Meritech tends to come in at the inflection where unit economics are proven.
Company and Leadership
Founded in 2023, Avoca is headquartered in New York with a Santa Barbara office. The company is a YC alum and hit unicorn status in under three years. Total funding now exceeds $125M.
Problem and Opportunity
The US home-services economy (HVAC, plumbing, electrical, pest control) is a $1T market run on landline phones, paper invoices, and dispatcher software from the 1990s. Calls go unanswered, jobs get mis-scheduled, and follow-ups slip. AI penetration today is under 1%.
Product and Technology
Avoca provides 24/7 AI voice agents that handle inbound call answering, appointment scheduling, custom marketing campaigns, and customer follow-ups for trades businesses. Customers include Turnpoint, 1-800-GOT-JUNK?, and Goettl. Partners include ServiceTitan, Nexstar, and Clover.
The ServiceTitan integration is the wedge. ServiceTitan is the dominant trades CRM, and a deep integration creates a one-click adoption path for tens of thousands of trades businesses. Voice agents are tuned for the actual quirks of trades intake: job pricing, dispatch logistics, customer-history retrieval. Generic voice-AI can't replicate this without comparable industry depth.
Use of Proceeds and Vision
The capital funds expansion into more service verticals beyond core trades, deeper ServiceTitan and dispatch-system integrations, and improvements to voice quality, latency, and edge-case handling. The pitch is to be the operating system for America's services economy.
Market Context
Vertical AI for SMBs and trades is one of 2026's hottest categories. Competitors include Rilla in sales coaching, Sameday in HVAC scheduling, Goodcall, and a wave of YC-backed voice-AI startups. The home services TAM is $500B+ in the US alone.
4. DeepInfra Lands $107M For Open-Source AI Inference Cloud
Deal Overview
- Stage: Series B
- Sector: AI infrastructure
- Geography: United States (Bay Area)
- Round size: $107M
- Closed: May 4, 2026
Investor Profile
500 Global and Georges Harik (ex-Google, co-founder of imo, prolific early-AI angel) co-led. Participants included A.Capital Ventures, Crescent Cove, Felicis, NVIDIA, Peak6, Samsung Next, Supermicro, and Upper90. NVIDIA and Supermicro show up as both investors and hardware partners, which matters for GPU allocation in a supply-constrained market.
Company and Leadership
DeepInfra was founded in 2022 in the Bay Area as a purpose-built cloud inference platform. Token throughput has grown 25x since the Series A, and revenue has tripled since the start of 2026.
Problem and Opportunity
Open-weight models (Llama, DeepSeek, Mistral, Qwen) are closing the quality gap with closed APIs. Developers and enterprises that don't want lock-in to OpenAI or Anthropic need a serving cloud that's reliable, cheap, and production-grade. API resellers running on top of hyperscaler clouds can't compete on price or quality control.
Product and Technology
DeepInfra runs a dedicated inference cloud serving 190+ open-source models, processing roughly 5 trillion tokens per week. It owns and operates its own GPU infrastructure rather than reselling hyperscaler capacity. Per-token pricing sits near the cheapest in the market.
The engineering moat is in batching, KV-cache management, speculative decoding, and quantization. Owning the GPU fleet end-to-end captures margin and quality control that resellers structurally can't.
Use of Proceeds and Vision
The round funds global compute capacity expansion, deeper developer tooling (eval, observability, fine-tuning), and support for the next generation of open-source and agentic models. The positioning is to become the AWS-equivalent for open-weight model serving.
Market Context
The open-source inference market is a four-way race: Together AI ($3.3B+ valuation), Fireworks AI, Anyscale, and DeepInfra. Goldman Sachs estimates inference TAM at $200B+ by 2030. Hyperscalers (Bedrock, Vertex, Azure AI) compete from above on bundled enterprise contracts; specialized clouds compete on $/token and developer experience.
5. Parallel Web Systems Raises $100M At $2B Valuation For Agentic Search
Deal Overview
- Stage: Series B
- Sector: AI (search infrastructure)
- Geography: United States (San Francisco Bay Area)
- Round size: $100M
- Valuation: $2B
Investor Profile
Sequoia led, with Kleiner Perkins, Index Ventures, and Khosla Ventures participating. Kleiner and Index led the prior Series A in November 2025 at a $740M valuation. The Series B closed just five months later at a $2B mark, a cadence that signals an aggressive land-grab in the agent infrastructure category.
Company and Leadership
Parallel Web Systems was founded in early 2024 by Parag Agrawal, former CEO of Twitter (2021 to 2022) and Twitter's CTO before that. Total raised is around $230M across seed, Series A, and Series B.
Problem and Opportunity
AI agents need to read the web, but human search engines aren't built for them. Google and Bing optimize for human SERPs: ten blue links, ads, ranking signals tuned for click-through. Agents need structured, deduplicated, citation-grade retrieval that they can act on programmatically.
Product and Technology
Parallel provides web search and research APIs purpose-built for AI agents. The product surface includes search, deep research, and tool APIs designed as the substrate agents call for fresh, accurate information. Customers include Clay, Harvey, Notion, and Opendoor; 100,000+ developers are on the platform.
The moat is a proprietary crawl, index, and ranking stack tuned for agentic recall and citation accuracy. That's a different optimization target than Google or Bing, and it's where the defensibility sits.
Use of Proceeds and Vision
Capital scales the crawl and index infrastructure, expands the developer platform and tool API surface, and grows the engineering team. The pitch is to be the AWS-equivalent for the agentic web, the layer every AI agent calls when it needs to read or research the internet.
Market Context
The AI-native search API market is exploding: Exa, Tavily, Perplexity API, Brave, You.com. Defensibility against Google and OpenAI native search is the open question. TAM is tied to the agent compute layer, which is projected to reach tens of billions as agentic workloads scale.
6. Moment Energy Secures $40M For Second-Life Battery Platform
Deal Overview
- Stage: Series B+
- Sector: Climate tech (energy storage)
- Geography: United States and Canada
- Round size: $40M
Investor Profile
The round was backed by Evok Innovations, Liberty Mutual Investments, W23 Global Fund, and Acario. The mix of climate-specialist VC, strategic insurer capital, and corporate venture suggests both financial and operational support for grid-scale storage deployments.
Company and Leadership
Moment Energy is building what it describes as North America's largest second-life battery platform. The company takes retired electric vehicle battery packs and repurposes them for stationary energy storage applications.
Problem and Opportunity
EV battery packs typically retire from vehicles with 70 to 80 percent of original capacity remaining. Disposal or recycling at that point wastes usable energy storage. Meanwhile, grid-scale and commercial stationary storage demand is outpacing new battery supply, and lithium prices are pressuring project economics.
Product and Technology
Moment Energy collects retired EV packs, tests and grades them, then assembles them into commercial and industrial energy storage systems. The economics work because second-life packs cost a fraction of new lithium-ion at comparable cycle life for stationary applications.
Use of Proceeds and Vision
The Series B+ scales North American deployment capacity, expands automotive OEM partnerships for retired-pack supply, and grows the manufacturing footprint. The vision is to build the dominant repurposing platform as the first wave of mass-market EVs reaches end-of-vehicle-life over the next five years.
Market Context
Second-life battery storage is forecast to hit 200+ GWh of cumulative deployments by 2030 as EV penetration matures. Competitors include B2U Storage Solutions, Connected Energy, and Smartville. The thesis depends on retired pack supply scaling fast enough to meet stationary demand, which is now starting to happen.
Lessons For Founders
- Owning the full stack still wins in regulated markets. Corgi's choice to be a licensed carrier rather than an MGA is what creates the data and unit-economics moat. The painful regulatory path is the moat.
- Speed of follow-on rounds is now a signal in itself. Corgi (4 months), Parallel (5 months), and Avoca's compressed cadence all show that when product-market fit is real, investors will price the next round before the last one finishes deploying. Plan for it.
- Vertical AI beats horizontal AI when the workflow is specific and the integration is hard. Avoca's ServiceTitan integration and DeepInfra's GPU-fleet engineering are both examples of moats that don't show up in a demo but compound in production.
- Infrastructure bets are getting funded at a scale that demands real conviction. Panthalassa's $140M for an unproven offshore compute thesis is only possible because the AI capacity crunch is now a board-level concern at every hyperscaler.
- Reuse and circular economics are no longer niche climate plays. Moment Energy's second-life storage thesis works because the math finally pencils out, and capital is showing up to scale it.
Get your round closed. Not just pitched.
A structured fundraising process matched to your stage and investor fit.
- Fundraising narrative and structure that holds up
- Support from strategy through investor conversations
- Built around your stage, model, and timeline