The Power to Choose Your Own Path

Index Ventures

Index Ventures just put $3.5B behind a 30-year thesis: great founders can come from anywhere, but the investors built to stay from the first check through the public markets remain a much shorter list.

The structure is precise. Index has a new $400M seed fund, a new $900M venture fund, and a $2.2B growth fund. That growth pool includes the firm's $1.5B 2024 vehicle plus a $700M upsizing. The result is $2B in new or incremental commitments and $3.5B in total investing capital, a distinction the headline alone can easily miss.

Index was founded in Geneva in 1996 by Neil Rimer, David Rimer, and Giuseppe Zocco. The firm has spent three decades turning geography from a constraint into a sourcing advantage, backing founders across Europe, Israel, and the U.S. before "global venture" became standard brochure copy.

The portfolio explains the model: Figma, Revolut, Roblox, Datadog, Adyen, Wiz, Robinhood, and Scale AI. Different markets, different cycles, and the same bet that the rarest thing in venture is not capital. It is conviction that survives the years before consensus arrives wearing a nicer jacket.

The timing is not subtle. NVCA and PitchBook say firms raised nearly as much in the first half of 2026 as they did during all of 2025, yet commitments remained concentrated among a small group of established managers. In 2025, the 10 largest U.S. funds captured 32.9% of venture fundraising. Capital returned. It did not return evenly.

Index is leaning into that imbalance. The seed fund creates more opportunities to meet founders before the market polishes the story. The venture fund expands early-stage firepower. The growth fund allows Index to keep supporting winners as later rounds become more capital-intensive and cap tables attract tourists.

AI is the obvious accelerant, but the strategy extends well beyond a single theme. Index points to infrastructure, cybersecurity, fintech, healthcare, productivity, and consumer markets where faster company formation can produce new category leaders. One partnership can price uncertainty at seed, build conviction at venture, and reinforce it through growth.

For founders, the useful question is not whether $3.5B sounds impressive. It does. The question is which investors can recognize a company before proof exists, then remain valuable when it becomes too large for hand-holding and too important for passive capital.

The new funds reinforce the firm's long-standing model of supporting companies across multiple stages instead of handing them off as they mature. The market is global, and the checks are larger, but the hardest part remains unchanged: recognizing conviction before everyone else calls it obvious.

Warren Equity Partners

Warren Equity Partners just closed Fund V with approximately $2.8B in commitments, and the number tells only half the story. The fund exceeded its $2.25B target in less than 12 months, reached its hard cap, and attracted more than 60 institutional investors from around the world.

That is institutional capital leaning into the machinery that keeps economies functioning: power and utilities, water and wastewater, transportation, waste, digital infrastructure, and buildings and facilities. The pipes, grids, networks, and service businesses everybody notices only after they stop working.

Warren Equity Partners has spent more than a decade building around that reality. Managing Partner and Co-founder Steven Wacaster and Partner and Co-founder Scott Bruckmann have kept the firm focused on companies that maintain, operate, and upgrade critical assets. The strategy favors recurring demand, fragmented markets, and operating businesses that cannot be replaced by a clever investor letter.

Fund V is roughly twice the size of Fund IV, which closed at more than $1.4B in 2023. Warren Equity Partners now reports $9.6B in pro forma regulatory AUM, 38 platform investments, more than 175 add-on acquisitions, and more than 215 total transactions since 2015. Scale expands reach, but it also raises the standard. More capital does not make good infrastructure businesses easier to find.

The first disclosed Fund V platform investment makes the thesis tangible. USG Water Solutions provides outsourced water-tank maintenance and asset-management services to municipal utilities. Nobody is building a consumer cult around water-tank compliance. The systems still need the work done, and neglect still sends the bill with interest.

The signal inside this close is straightforward. Limited partners are backing a sourcing and operating model built for markets where demand is recurring, assets are aging, and consolidation opportunities remain. Warren Equity Partners says its resources are increasingly supported by a proprietary in-house AI platform, but the test remains physical: finding durable companies and helping them execute.

Fund V gives Warren Equity Partners more room to invest across North America and Europe, not permission to spray capital. The next chapter will be measured by how effectively Steven Wacaster, Scott Bruckmann, Henrik Dahlback, and the broader team convert institutional demand into stronger infrastructure businesses.

The close reinforces a broader lesson. When capital becomes more selective, a narrowly defined thesis supported by years of operating evidence can still attract a very large room.

Foundational Industries

Foundational Industries just raised a $25M seed round to build something venture capital rarely funds casually: factories.

Not another dashboard for the factory. Foundational Industries is building AI-native, autonomous factories where design ingestion, process planning, robotic execution, inspection, scheduling, quality, and supply-chain operations run through a single operating layer.

Most industrial automation arrives as an upgrade to a system that was never designed to learn. Foundational Industries wants to start with the intelligence layer, then build the physical system around it. The bet is that every production run improves the model, and every new facility benefits from what the last one learned.

BoxGroup and Zigg led the seed investment, with Abstract, Adverb Ventures, Buckley Ventures, and Offline Ventures participating. They are backing an operating model that owns both the factory and the software while remaining accountable for qualified production at scale.

Founder and CEO Jonathan Winer brings a career at the intersection of capital, infrastructure, and physical systems. Foundational Industries says Jonathan Winer has spent 25 years building technology-enabled infrastructure, deployed more than $3B of capital, and helped create physical AI companies at Alphabet's Sidewalk Infrastructure Partners. Capital can finance capacity, but it cannot manufacture the part when the domestic factory does not exist.

The first facility is planned for the end of 2026. Its initial focus will be manufacturing capacity for AI data-center equipment, where faster silicon keeps dragging a much heavier physical tail behind it. Power, cooling, interconnects, and supporting infrastructure still have to be designed, produced, inspected, and delivered in the real world.

The AI economy does not stop at models and GPUs. It extends into cooling systems, connectors, robotic cells, quality loops, suppliers, and every other part of production that nobody notices until a shortage stops the plan.

Foundational Industries is making a sharper bet: America does not win the next manufacturing era by rebuilding yesterday's factory with more expensive labor. It wins by changing the production architecture and turning software, robotics, data, and physical operations into a single system that improves across an entire network.

The financing gives Foundational Industries the opportunity to prove that autonomy can move from factory thesis to qualified production that customers trust.

June.ai

June.ai just came out of stealth with a $20M pre-seed round and a very grown-up thesis: the hard part of enterprise AI is not building another agent. It is getting that agent to survive contact with the systems already running the business.

Time Ventures led the round, with backing reported from Michael Dell, Aaron Levie, and George Kurtz. The more interesting signal is what they are underwriting. Enterprises do not lack access to models. They lack a clean path from an impressive demo to a production system that can navigate legacy workflows, fragmented data, approvals, security, and organizational politics.

June AI is built for that messy middle. The platform maps how work moves through systems such as Salesforce, ServiceNow, Databricks, Workday, Oracle, Snowflake, SAP, and Microsoft. It identifies where AI can create value, lays out the implementation steps, and helps build approved changes.

"We built an agent" and "the company can rely on it" are still two different sentences. Large organizations have years of business logic buried inside software configurations, handoffs, and tribal knowledge. June AI is betting the next enterprise AI winner will understand the machinery around an answer well enough to make the change stick.

The founding team has seen this movie from both sides. CEO Efrat Rapoport, President Barak Goldstein, Chief Architect Ohad Hen, and CTO Idan Tsitiat previously built Bonobo AI. Salesforce acquired Bonobo AI in 2019, and the team spent the following years building AI inside one of the platforms now sitting at the center of the implementation challenge.

TechCrunch reported an early example at CMG, where Chief Strategy Officer Paul Akinmade was trying to move a Salesforce-related agent initiative forward after weeks of stalled progress. June AI gave the team a clearer view of where agents could be deployed and how to do it safely. One customer example is not a market verdict, but it is exactly the kind of evidence enterprise buyers will demand.

The $20M does not prove June AI has solved enterprise deployment. It gives the company room to determine whether its approach is repeatable across systems, industries, security models, and teams. The valuation, audited growth metrics, and a detailed use-of-funds plan were not disclosed, so the real scoreboard starts now.

June AI is not pitching AI as magic dust. It is attacking the expensive part everyone discovers after the demo, when software has to understand the business before the business can trust the software.

Inocras

Inocras Inc. just closed a $31M oversubscribed Series B-3, bringing total funding to approximately $100M and putting a serious market question in front of precision oncology: what happens when whole-genome analysis starts behaving like clinical infrastructure? The next proving ground is the United States.

Inocras says its sequencing solutions are used by more than 100 cancer institutions across Asia and have supported thousands of patient cases, with a presence in Hong Kong and approximately 30 hospitals in South Korea. Those are company-reported figures, but they provide meaningful operating evidence behind the company's expansion.

CancerVision profiles tumors across the whole genome, while MRDVision uses a tumor-informed approach to detect molecular residual disease. Proprietary bioinformatics and automation turn genomic complexity into insights clinicians and researchers can use. Generating more sequencing data is easy to admire. Making it useful on a hospital's timeline is where the work gets expensive.

Strategic investors NDS Corporation and Aimed Bio joined new backers IMM Investment, Korea Investment & Securities, LoftyRock Investment, DT& Investment, Woori Investment & Securities, and Shinhan Securities. DSC Investment, Dunamu & Partners, and InterVest also returned. No lead investor was disclosed, so no spotlight gets assigned by guesswork.

CEO Jehee Suh is leading the U.S. expansion alongside co-founders Young Seok Ju and Jeong Seok Lee, and CTO Sangmoon Lee. Founded in 2020, Inocras operates a CLIA-certified, CAP-accredited laboratory. Working with Broad Institute researchers, the company analyzed more than 8,000 tumor-normal whole genomes across more than 30 cancer types and more than 250M harmonized variant calls.

That scale matters because whole-genome sequencing does not win by producing a larger file. It wins when the laboratory, analytics, evidence, and clinical workflow align closely enough to make the information usable. Hospitals ask harder questions than demos answer: Is it repeatable? Can it be interpreted? Does it integrate? Does the evidence hold up in real patient care?

This financing gives Inocras more room to answer those questions. The capital will support U.S. commercial and operational infrastructure, hospital adoption, clinical expansion, and the evidence needed to move whole-genome analysis closer to routine cancer care. The next test is whether an Asia-tested model can meet the demands of a U.S. market that will scrutinize every claim. Can comprehensive genomics become ordinary enough to matter every day?

Dash Bio

Dash Bio just raised a $30M Series A to make one of drug development's least glamorous jobs move much faster. Oak HC/FT led the round, with Freestyle Capital and Swift Ventures returning, bringing the company's total funding to $47.5M.

Drug discovery keeps getting faster. AI can propose candidates, new modalities continue expanding the frontier, and biology has no shortage of ambition. Then those programs reach development infrastructure built around manual work, paper records, opaque pricing, and long queues.

Dash Bio is attacking that gap through bioanalysis, the laboratory work that turns preclinical and clinical samples into evidence about safety and efficacy. The company has built a 24,000-square-foot GLP-compliant laboratory around robotics, proprietary software, digital traceability, and productized assay workflows spanning ELISA, MSD, qPCR, ddPCR, and LC-MS.

This is not software wrapped around the same service model. Dash is rebuilding how the work moves. The company publishes fixed pricing, automates critical steps, and says it can return preclinical sample results in days rather than months. That shortens decisions that would otherwise wait on laboratory data.

The traction is beginning to match the architecture. Dash says it has signed contracts with more than 30 customers since beginning go-to-market efforts about nine months ago, including four of the world's 15 largest pharmaceutical companies. Those are company-reported figures, but they sharpen the investor thesis: buyers are not shopping for a clever laboratory demo. They are testing a different way to run development work.

Oak HC/FT is betting that scientific infrastructure can become a platform rather than a staffing equation. That aligns with the experience of co-founders Dave Johnson, CEO, Ander Tallett, COO, and Ely Porter, CSO, who built the company after helping develop digital and automation infrastructure at Moderna. Porter adds the scientific depth needed to ensure automation holds up across real assays and regulated workflows.

The $30M will support commercial expansion, broader assay offerings, additional laboratory capacity, entry into new categories, and continued development of Dash's proprietary software platform. The next stage is proving that this operating system can scale without sacrificing the quality and predictability that attracted customers in the first place.

The broader signal reaches beyond Dash Bio itself. Modern science eventually demands modern infrastructure, and the cost of waiting keeps rising.

Horizon3.ai

Horizon3.ai just raised a $250M Series E at a valuation above $2B, just over a year after a $100M Series D valued the company at $650M. That is not a gentle step up. It is the market putting a larger number behind a specific bet: security teams need proof of what attackers can exploit, not another dashboard explaining what might go wrong.

NightDragon and NEA co-led the oversubscribed round. Seven new investors joined five returning backers. NightDragon founder and CEO Dave DeWalt and Managing Director Morgan Kyauk are joining Horizon3's board.

The product behind that conviction is NodeZero. It autonomously attacks an organization's production environment, finds exploitable paths across credentials, identity, cloud, endpoints, and infrastructure, then guides remediation and verifies the fix. A vulnerability list tells a team what exists. An attack path shows what can actually be chained into damage.

Horizon3 says NodeZero has completed 310,000 production-safe tests, the company protects more than 7,000 organizations, and annual recurring revenue grew 120% year over year. Those are company-reported figures, but they explain why this round is about scaling an operating model rather than rescuing a science project.

The capital will expand sales, marketing, and channel operations across enterprise, mid-market, and federal customers. Horizon3 also plans to enter Singapore and Australia, deepen its EMEA presence, and connect AI attackers with AI defenders, including autonomous blue-team agents that can remediate findings from NodeZero tests.

That is the broader signal. AI is compressing the time between discovering a weakness and exploiting it. Annual penetration tests and compliance theater were already strained by modern infrastructure. Machine-speed attacks make that cadence look even more outdated. The winning security stack will not simply generate more alerts. It will continuously test whether defenses work, prioritize what is exploitable, fix it, and verify that the fix holds.

CEO and co-founder Snehal Antani frames cybersecurity as algorithms fighting algorithms, with humans handling the exceptions. Co-founder Anthony Pillitiere and the founding team built Horizon3 from experience across U.S. Special Operations, national security, and enterprise technology. Six years later, the company is selling a measurable answer to a question every CISO faces: can our controls stop a real attack?

NightDragon and NEA are backing a shift from assumed security to continuously verified security. The pace of modern attacks is making that transition increasingly difficult to ignore.

Marquee

Marquee raised a $4M seed round, bringing total funding to $6.5M, because professional sports has reached a strange point: the data is everywhere, but the decision is still trapped across a dozen systems and spreadsheets.

The company is building an AI-native decision layer for clubs, federations, and recruitment teams. It connects external data providers with a club's internal models, institutional knowledge, financial constraints, and sporting identity. Then it turns that complexity into ranked shortlists, 360-degree scouting reports, squad-fit analysis, tactical breakdowns, and market signals delivered in natural language.

Marquee says clubs typically use only 4% to 15% of the information available to them. The company also says its platform can streamline roughly 70% of existing recruitment workflows, compressing analysis that once required weeks into minutes. Those are company-reported figures, but the operating problem behind them is familiar to anyone who has watched tools multiply faster than good decisions.

The platform is already live with more than 20 clubs and federations across multiple countries, including Tier 1 clubs in the Premier League and Serie A, as well as organizations in MLS. Partnerships with Gradient Sports and SkillCorner combine physical, positional, event, financial, and internal data without forcing analysts to become systems integrators.

The founding team combines deep domain interest with technical expertise. Dean Bracha is CEO, Dror Rosenfeld is CPO, Jonathan Hazut is CTO, and Tal Darchi is Chief Data Scientist. They met through an Arsenal supporters community, a very football way to begin building enterprise software: debate players, then build better machinery for the evidence.

Axel Springer SE, Welltech Ventures, APEX Capital, AnD Ventures, 97212 Ventures, and Jefferson Capital Partners are backing the next phase. Marquee plans to accelerate R&D, expand its AI, engineering, and data science teams, open commercial offices in the U.S. and UK, and move beyond football. Basketball development is already underway.

The broader signal extends beyond scouting. Vertical AI earns its place when it understands the workflow, preserves expert judgment, and helps operators make better decisions while those decisions still matter. Marquee is using this funding to prove that model in one of the world's most emotional, expensive, and unforgiving talent markets.

Mate Security

Mate Security just raised a $35M Series A, and the useful part of the story is not that investors found another cybersecurity company with AI in the sentence. It is what they are betting security teams will need when attacks move faster than humans can investigate them.

Canaan Partners led the round, with Insight Partners, Team8, and M12, Microsoft's Venture Fund, participating. The financing arrives just eight months after Mate emerged from stealth with a $15.5M seed, pushing total disclosed funding beyond $50M. That pace earns attention. It does not automatically earn the right to declare victory or product-market fit.

Mate's thesis is more specific. AI agents do not become trustworthy because the model gets louder, faster, or more expensive. They become useful when they understand the organization they are defending: its people, systems, policies, history, exceptions, and business context.

That is the role of Mate's Security Context Graph. The company says it gives agents a shared knowledge layer for detection, triage, investigations, supervised response, and threat hunting, turning isolated alerts into decisions grounded in how an enterprise operates.

The architecture exposes an uncomfortable reality behind the AI security rush. Bolting a general-purpose model onto an old SOC workflow may improve the interface while leaving the underlying operating problem untouched. Speed without context can produce faster uncertainty.

Mate reports revenue growth of more than 500% since Q3 2025, adoption among Fortune 500 enterprises, and nearly 100% wins in proofs of value and competitive evaluations. Those are company-reported figures, not independently audited, but they help explain why Insight Partners and Team8 returned while Canaan and M12 joined the round.

Co-founders Asaf Wiener, CEO, Oren Saban, CPO, and Guy Pergal, CTO, are building from experience across Wiz, Microsoft, Axonius, and Apex. The team has lived inside the systems it now wants to redesign, which matters in a market where a polished demo can hide operational reality.

The $35M will support platform expansion, growing customer demand, and expansion into new markets and product categories. The test is whether Mate can turn a compelling context thesis into durable deployments, measurable security outcomes, and AI economics that continue working after the demo ends.

The sharper signal belongs to operators: enterprise AI will not be trusted because it can act. It will be trusted when it can explain what it knows, respect what it should not touch, and make better decisions inside the messy reality of the business.

Quercus Biosolutions

Quercus Biosolutions just closed an oversubscribed $5M seed round. The real story is a direct bet that crop protection can borrow computational tools from drug discovery before resistant weeds keep narrowing the farmer's options.

Founded in 2024 by Jonathan Lightner, PhD, and Matt Crisp, Quercus is tackling a familiar problem: biological crop protection has often struggled to deliver the consistency, cost, and field performance growers expect. Its answer is generative AI-designed mini-proteins built for validated biological targets in weeds and other pests.

Instead of screening only what nature already produced, Quercus works with Ordaos Bio's technology to create candidates from scratch. The proteins are designed to bind biological targets differently from conventional small-molecule herbicides, creating a potential path around some forms of evolved chemical resistance. The commercial test is field performance, regulatory approval, manufacturing economics, and agriculture's wonderfully patient calendar.

Quercus has five active programs and multiple field trials underway. The team is designing with production constraints in mind, using computational testing, cell-free expression, and microbial systems such as E. coli and Pichia, while third parties evaluate manufacturing costs and scale-up. Translation is where elegant science meets regulators and a field that does not care how impressive the pitch deck looked.

The new capital will expand the herbicide pipeline, add new biological targets, strengthen the company's IP portfolio, begin regulatory work, and move at least one product toward late-stage development. Quercus also plans to explore fungi, insects, and other pests, giving the platform more than one shot on goal without pretending every shot is already a product.

Serra Ventures led the round. Starshot Capital, Grantham Foundation for the Protection of the Environment, Flyover Capital, Rootstock Capital, Iowa Corn Opportunities Fund, AgriCap Fund, Kentucky Corn Plus, AgVentures Alliance, and Ag Startup Engine also participated. That cap table combines agriculture, grower-backed capital, climate, and technology investors, fitting a company whose science will require both technical patience and agricultural distribution.

Jonathan Lightner serves as CEO, alongside co-founder Matt Crisp. The lesson for operators is straightforward: AI becomes valuable when it changes the economics and speed of a specific workflow, not when it simply decorates the vocabulary. For Quercus, $5M now funds the hardest kind of proof: whether designed proteins can move from promising programs to credible crop-protection products.

Harmony

Harmony just raised $34M in seed funding to attack a very expensive enterprise habit: making employees route simple requests through portals, queues, forms, and people who already have better work waiting.

Lightspeed Venture Partners led the round, with Hitachi Ventures, Fin Capital, Mercer Ventures, and Operator Partners participating. Named angel investors include Assaf Rappaport of Wiz and Ofir Ehrlich of Eon.io.

Co-founder and CEO Nitzan Shapira and co-founder and CTO Ran Ribenzaft previously built Epsagon, the cloud observability company Cisco acquired in 2021 in a deal reported at $500M. Harmony is their second company together, this time focused on the operating layer every employee touches.

Harmony places AI agents inside Slack and Microsoft Teams so employees can request application access, equipment, policy guidance, payroll support, onboarding assistance, or technical help without learning which back-office maze owns the answer. Its context graph connects identity, devices, applications, permissions, and work history, then enables agents to act across the systems behind each request.

The important distinction is not chatbot versus portal. It is resolution versus routing. Traditional enterprise service software is very good at documenting who is waiting on whom. Harmony wants the software to complete more of the work before a ticket ever becomes someone else's problem.

The company says it has more than 100 prebuilt agents and customers including n8n, eToro, Cyera, and KITH. Harmony also reports that deployments have deflected 48% of support requests within two weeks and more than 75% within three months, with some organizations exceeding 90% autonomous resolution. Those are company-reported results, but the useful metric is work completed, not AI features admired in a demo.

The $34M will support product development, hiring, deeper enterprise integrations, and expansion across infrastructure, enterprise applications, cybersecurity, and ERP. That scope matters because internal service is rarely one department's problem. Employees see one company. The software stack sees a scavenger hunt.

Lightspeed is backing a shift from tickets and queues toward AI agents that resolve requests. It is a compelling wedge into enterprise AI: start where repetitive work is measurable, expensive, and painfully familiar.

The broader signal is that experienced founders and investors are betting the next enterprise platform is not another place to manage work. It is a system that quietly gets more of it done.

Microporous

Microporous secured the final financing needed to move its Danville, Virginia, battery-separator plant into construction. This is a manufacturer with more than 90 years of operating history putting capital behind a critical piece of the domestic battery supply chain.

The financing package is led by Elda River Capital Management and Eagle Point Credit Management, with backing from owner Trent Capital Partners. The April 30 announcement does not disclose the financing amount, pricing, maturity, or covenants. The project figures are significant, but they should not be confused with the size of the credit facility.

Virginia previously announced a broader $1.35B Microporous investment expected to create 2,015 jobs across the planned Pittsylvania County buildout. Separately, the U.S. Department of Energy described $100M in cost-shared funding toward a $525M Phase I project, alongside a $425M private cost share. Those figures describe the project's scale, not the lenders' commitments.

The Danville facility is expected to produce coated lithium-ion battery separators and support 65 GWh of annual battery production. Microporous has also said the plant is designed for initial output of 600M square meters per year. This is industrial capacity, not a software feature that ships after a weekend sprint.

Microporous brings substantial operating history to that construction risk. Its roots trace to 1934, the Microporous name dates to 1991, and the company already manufactures in Tennessee and Austria. It built its foundation in lead-acid separators before expanding into coated polyethylene separators for lithium-ion batteries.

CEO John M. Reeves framed the financing as the point where planning becomes execution. President Doug Rich and the Microporous team now have to translate capital into commissioning, quality control, customer qualification, and dependable output. In advanced manufacturing, money opens the gate. Process discipline determines whether the line runs.

Trent Capital Partners provides the ownership foundation. Elda River brings energy and infrastructure capital. Eagle Point contributes an infrastructure-credit strategy led by Jennifer Powers. Together, they are backing a component that sits between the anode and cathode while carrying an outsized share of battery safety and performance.

Credit financing rarely receives the attention of a priced equity round, but factories care about steel, coating lines, utilities, trained operators, yield, and qualified production. The next signal is no longer the financing package. It is execution in Danville.

Fly.io

Fly.io just raised a $25M Series D, and the interesting part is not another cloud company collecting another check. It is the operating decision behind the financing: Fly.io is going all-in on computers built for agents.

Dell Technologies Capital and Intel Capital co-led the round, with Andreessen Horowitz, EQT, Geodesic, and Y Combinator participating. Martin Casado of Andreessen Horowitz and Daniel Docter of Dell Technologies Capital are joining the board, giving the financing more infrastructure weight than a typical growth-capital announcement.

Fly.io has spent years making real compute approachable for developers. Fly Machines are fast-launching Firecracker microVMs with persistent storage, private networking, global placement, and direct lifecycle control. That foundation now matters for a different user: software agents that need to maintain state, reach other systems, survive restarts, and continue working after the demo applause fades.

The company calls its new focus "computers for agents," with Sprites moving from a small internal project to the center of the business. A disposable sandbox is useful for a task, but an agent expected to perform production work needs a durable computer, an address, persistent storage, and secure connections to the tools around it. Temporary compute can run code. Persistent compute can carry responsibility.

Fly.io says more than 37,000 customers build on the platform, including more than 8,000 agent-native customers. The company also reports nearly 12x revenue growth over the past 12 months among its largest agent-native customers, while agent-native companies now account for roughly two-thirds of revenue from that group. Those are company-reported figures, but they explain why this strategy is arriving with capital instead of a speculative memo.

The round also comes with a leadership transition. Scott Johnston, former Docker CEO, is joining Fly.io as CEO and a board member. Founder Kurt Mackey is moving into an advisory role while remaining on the board, and CTO Jérôme Gravel-Niquet continues leading the technical organization. Fly.io gains an operator who has scaled developer infrastructure while preserving the technical continuity that brought the company to this point. That is a meaningful operating decision, not a branding exercise.

The sharper question is whether persistent computers for agents become a distinct market or simply the direction cloud infrastructure was always heading once software began operating itself.

CoreMap

CoreMap's latest financing is a reminder that medical-device capital is not buying a prettier pitch deck. It is buying time to prove that a hard clinical problem can be measured, mapped, and treated with greater precision.

A June 24 SEC filing shows CoreMap sold $35.5M in equity toward a $45M offering. The filing does not identify the round as a Series C, name the current investors, or disclose a valuation, so those details remain where unsupported information belongs: outside the story. What the record does show is substantial capital moving into a company trying to make persistent atrial fibrillation less mysterious inside the electrophysiology lab.

CoreMap was founded in 2016 by Sarah Kalil, President and CEO, and Peter Spector, M.D., Chief Scientific Officer and inventor. Their thesis is blunt: physicians cannot tailor an ablation strategy if the electrical behavior driving AF remains difficult to see. The company's investigational mapping system combines proprietary microelectrodes with high-resolution signal capture and software designed to identify spatially constrained drivers of the arrhythmia.

This is the medical-device version of attacking the measurement problem before pretending the treatment problem is solved. It is less glamorous than promising a cure in three slides, but it is often where durable clinical platforms begin.

The next proof point is INvENI, CoreMap's recruiting global, multi-site feasibility study. ClinicalTrials.gov lists an estimated 245 participants and a December 2026 completion target. The program is evaluating whether CoreMap-guided mapping and tailored ablation can demonstrate safety and effectiveness in persistent and long-standing persistent AF. U.S. Food and Drug Administration IDE approval in 2025 expanded that work into the United States.

For founders, there is a useful financing lesson here. A $35.5M equity sale does not validate a clinical outcome. It validates the opportunity to keep earning one. In medtech, capital pays for protocol execution, regulatory discipline, manufacturing readiness, and the long stretch between a compelling signal and a product physicians can trust.

For investors, the wager is equally clear. Persistent AF is a serious treatment challenge, current mapping remains imperfect, and better electrical visibility could make ablation more patient-specific. The opportunity is large, but the burden of proof is larger. That is exactly why the funding matters.

Sarah Kalil, Peter Spector, M.D., and the CoreMap team have aligned the science, study design, and capital behind the same objective. The company has more runway. Now the data gets the final word.