What You Control

trutility

trutility launched with an anchor investment from Distributed Sun, but this is not the usual financing story with a neat round label and a victory-lap valuation. No amount was disclosed. No Series A badge was pinned to the release. The currency here is operating infrastructure.

Distributed Sun is contributing a GW-scale front-of-meter development portfolio and productized intellectual property. truCurrent, its behind-the-meter subsidiary, brings the operating engine. trutility says it owns the contributed assets, customer relationships, and operating system.

That structure matters because electricity demand is moving faster than the machinery built to serve it. The U.S. Energy Information Administration says power demand is rising, with data centers a major driver. Interconnection queues keep turning ready to build into a phrase that needs an asterisk.

trutility is aiming at that gap with an integrated platform spanning battery storage, community solar, distributed generation, and microgrids on both sides of the meter. The pitch is not another dashboard that tells operators the grid is congested. It is to design, develop, own, and operate the assets that turn a power problem into usable capacity.

The company says its portfolio includes 24 front-of-meter battery sites, with that count expected to double this year, across 14 utility rate territories. It also points to an EV fleet-charging microgrid energized four months after groundbreaking through a PG&E Flex Connect partnership. Those are company-reported operating claims, not independent guarantees, but they show what trutility wants to be judged on: time to power and asset performance.

Chase Weir, CEO of Distributed Sun, frames electricity as the base currency of the economy. Jeff Weiss, Distributed Sun's executive chair, brings the investor-side discipline. Their wager is that energy infrastructure can be treated less like a collection of disconnected projects and more like a compounding operating platform.

There is a sharp business lesson in that. AI infrastructure is not only a chip, model, or data-center story. It is also a permitting, interconnection, storage, financing, and execution story. The companies that make power available sooner may create as much leverage as the companies consuming it.

Vals AI

Vals AI just raised a $40M Series A at a $400M valuation to answer a question the AI market keeps trying to grade on a curve: can these models actually do the work?

The round was led by a16z, with existing investors 8VC, Pear VC, and Bloomberg Beta returning alongside new investors HRT Ventures and Next Ladder Ventures. That syndicate is not betting on another leaderboard. It is betting that independent measurement becomes critical infrastructure when every model vendor arrives with its own report card and a suspicious amount of confidence.

Co-founders Rayan Krishnan, CEO, and Langston Nashold, CTO, built Vals AI around real professional tasks instead of academic theater. The company works with domain experts to test models across finance, law, healthcare, coding, and other fields, then keeps key test sets private to reduce leakage, contamination, and gaming. A model does not get credit for knowing the vocabulary. It has to produce useful work.

That distinction gets more valuable as AI moves from answering questions to operating inside business workflows for hours. Picking the wrong model is no longer a funny demo failure. It can mean wasted compute, broken processes, bad analysis, and customers wondering why the robot needs adult supervision.

Vals AI says its results have been cited in model cards from OpenAI, Anthropic, Google, Meta, and xAI. The company also reports revenue grew 8x compared with all of 2025, while its customer base doubled and its team tripled in six months. Those are company-reported numbers, but they show why investors see a market forming around the referee, not only the players.

The product expansion matters too. Vals Smith is now generally available, turning merged pull requests from a GitHub repository into coding benchmarks with hidden tests and reproducible setups. Vals also introduced frontier-risk benchmarks and a broader Vals Index. Translation: buyers can test models against the work they need done, not the exam every lab already studied.

The business takeaway is simple. As AI capability becomes cheaper and more abundant, trusted evidence becomes more valuable. Vendors know more than buyers, benchmarks expire, and polished demos hide the gap between capability and reliability. Markets eventually pay someone to make that gap visible.

Velaura AI

Velaura AI just raised $110M in Series A financing and crossed a valuation of more than $1B. Seligman Ventures led the round, Capricorn Investment Group joined as a new investor, and Mayfield, Maverick Silicon, MARA, Premji Invest, Samsung Catalyst Fund, and StepStone Group returned to the table.

AI keeps getting discussed like compute appears by magic, right up until the power bill, grid queue, cooling load, and thermal envelope walk into the meeting. Then the future suddenly needs an electrician.

Velaura AI is attacking that problem inside the silicon. Its Titan Core platform combines proprietary digital chip IP, low-voltage libraries, and design tools intended to help AI accelerators do more work per watt. The company reports a 2-4x improvement in performance per watt for mathematical operations and says the underlying technology has been deployed in more than 30M production ASICs.

The commercial model is just as interesting. Reuters reports that Velaura AI charges an upfront technology fee plus a royalty tied to a share of the power savings customers achieve. That turns efficiency from an engineering footnote into a line item both sides can measure. If the chip saves real money, Velaura AI participates in the value it created.

Rajiv Khemani told Reuters the company is engaged with three of the four largest cloud providers as potential customers, though none were named. The distinction matters. Engagement is not deployment, but it does put Velaura AI in the rooms where power budgets, accelerator roadmaps, and data-center economics get decided.

The market timing is not subtle. The IEA expects data-center electricity demand to roughly double from 485 TWh in 2025 to 950 TWh in 2030, while electricity use from AI-focused facilities is projected to triple. Everyone wants more compute. The grid keeps asking where they plan to plug it in.

The new capital will accelerate product development and commercialization while expanding engineering and customer-facing teams. That gives co-founder and CEO Rajiv Khemani, co-founder and Chief Development Officer Manu Gulati, co-founder and President, Strategy & GTM Sanjay Gupta, and President, Products YJ Kim more room to turn semiconductor expertise into deployed infrastructure.

Etched

Etched just raised $700M at a $21B valuation. The investor leading the round is also the first customer putting its hardware to work.

Jane Street led the financing, joined by Kleiner Perkins, Sequoia, Andreessen Horowitz, Tiger Global, Bain Capital Ventures, Neo, Primary, Stripes, Positive Sum, and Blackstone. Etched says it has now raised $1.9B. The company did not disclose a stage for this round.

Etched shipped its first rack to Jane Street in July, and the trading firm says it is deploying the system into its workloads. Jane Street tested the chip, became the first disclosed customer, then led the next financing. That sequence carries more weight than a logo crowded onto a cap table.

Etched is building frontier inference clusters, not selling a magic chip and hoping the rest of the data center figures itself out. The company co-designs silicon, memory, interconnects, racks, cooling, software, and manufacturing around the economics of running AI models. Low Voltage Inference targets compute density within a power envelope. Cluster Scale Memory creates a shared, low-latency memory pool across the system.

Etched says its systems are running large mixture-of-experts models and non-transformer designs. It also reports more than $1B in customer contracts across AI companies and cloud providers. Contracts are not revenue, and one rack is not mass production. Both matter when the hard part of semiconductors begins after the deck, the tapeout, and the first applause.

That hard part is now the story. Etched reported first-pass A0 silicon on TSMC's N4P process, has more than 400 people, opened an 80,000-square-foot, 10 MW facility in Milpitas, and is building three hardware generations in parallel. Less than one month earlier, Sequoia led a $300M Series C at a $10.3B valuation. The valuation doubled again. Factory physics did not.

Gavin Uberti, Robert Wachen, and Chris Zhu founded Etched in 2022 after leaving Harvard. Uberti is CEO, Wachen is President, and Mark Ross is CTO. Their bet was that inference would become the economic center of AI, demanding purpose-built systems rather than another incremental patch on general-purpose hardware.

The market is now funding that bet like an infrastructure race. The next proof will not be another valuation. It will be racks that ship, workloads that stay, and performance that survives outside Etched's lab.

Space

Space just raised $2.4M in Pre-Seed funding to give computers infinite room without downloading, duplicating, and babysitting every file. The round was led by a16z Speedrun, with Golden Ventures, Northside Ventures, and angels from Parsec, Sentry, Stan, Superwhisper, and Modem participating. That mix is less about storage capacity and more about where modern work gets stuck.

Space, founded by Matthew Ao, Arihant Bapna, and Jason Zhao, is building a distributed filesystem that mounts cloud storage as a native drive. Files look local to Finder and the applications people already use, but Space streams only the byte ranges requested at that moment. The whole file does not need to arrive before the work can begin.

That matters when a video project is larger than the laptop opening it, an architecture team is moving massive models, or an AI agent needs one slice of a dataset instead of another ingestion ceremony.

Cloud storage solved capacity. It did not solve movement. Teams still upload, download, sync, index, copy, compress, and then wonder why the supposedly frictionless cloud feels like a warehouse with one loading dock.

Space is attacking that access layer. Creative tools, code editors, CAD software, and agents can operate against the same live data instead of forcing every workflow into a separate connector.

The company says its private beta has onboarded about 100 users and teams, while its audience has grown beyond 80,000 across platforms. Those are early, company-reported signals, but they fit the founding problem. Jason Zhao dealt with dozens of terabytes of video; the founders later watched their previous company move terabytes of footage each month. This was not a whiteboard pain point. It was the tax bill from their own workflow.

The sharpest part of the thesis is not more cloud. It is that a computer can become a window into data instead of the place all data must physically wait. Humans and software agents then share the same filesystem primitive, opening only what they need, when they need it.

That creates a credible wedge into video, marketing, architecture, engineering, and construction, followed by the heavier frontier: training data, computer vision, media infrastructure, and enterprise systems where local disks lose the argument before the first file transfer starts.

Ascensus

Ascensus is entering a new ownership chapter with two private equity firms that already know the business. Stone Point Capital and Genstar Capital are each investing new capital and will hold equal stakes. GIC, an investor since 2019, will remain invested. Financial terms were not disclosed.

The ownership history adds context. Genstar was a lead investor from 2015 through 2021, and Stone Point joined the ownership group in 2021. The two firms will combine that experience under joint governance once the deal closes.

Ascensus has grown into a substantial part of the U.S. savings infrastructure. As of August 3, 2026, the company supported more than 16 million savers and oversaw more than $1.3 trillion in assets under administration across workplace retirement, education, disability, health, and emergency savings programs.

That $1.3 trillion figure describes assets under administration. It is not the investment amount or a company valuation, both of which remain undisclosed.

The transaction follows Ascensus's August acquisition of AmericanTCS, which added trust and custody services, pooled employer plan capabilities, fiduciary support, and administration technology. Together, the announcements show how Ascensus is broadening its infrastructure for institutions and savers.

CEO Nick Good said the new partnership will support continued investment in technology and artificial intelligence, client service and experience, and broader capabilities. Current leadership, client relationships, and the company's service model are expected to continue.

Stone Point's Fayez Muhtadie cited Ascensus's scale. Genstar's Tony Salewski and Sid Ramakrishnan emphasized wider access to tax-advantaged savings.

For clients, continuity may be the most important near-term message. Savings administration is regulated, long-duration work. New ownership has to preserve reliable operations while creating room for improvement.

For the market, the deal shows continued investor interest in scaled financial infrastructure. Ascensus sits behind programs used to save for retirement, education, health needs, and disability-related expenses. Better technology and service could therefore reach well beyond the company itself.

The operational test comes next. The transaction is expected to close in the coming months, subject to customary regulatory approvals and closing conditions. Stone Point and Genstar will assume joint governance at closing.

Boom

Boom just raised a $15M Series A, and the interesting part is not that another proptech company found capital. It is what S3 Ventures is betting rental software should become.

The Austin company started in 2020 with BoomReport, a way to turn rent payments into credit-building data. It then moved deeper into the machinery with BoomScreen, giving property managers configurable applicant screening and underwriting. Now Boom is launching BoomCRM, an agentic leasing and touring product built on top of that same infrastructure.

That order matters. Most leasing software began as a ledger or communication tool, then added screening, automation, and AI wherever the seams would tolerate them. Boom began with regulated data, compliance, and underwriting, then built the customer-facing workflow above it. The result is less chatbot with a property tab and more one operating layer from first inquiry through approved resident.

S3 Ventures led the round. Mischief VC participated alongside repeat investors Starting Line VC, Company Ventures, and Gilgamesh Ventures.

Investors are following evidence, not a clean pitch deck wearing loafers. CEO and co-founder Rob Whiting says Boom now serves more than 500 operators covering more than 500,000 units. The company reports 8x year-over-year revenue growth into the double-digit millions.

Those numbers are company-reported, but the market logic is clear. Single-family rental and manufactured-housing operators manage scattered properties, regional rules, touring logistics, and fraud risk. A generic multifamily CRM was never designed for that mess.

BoomCRM can answer calls, qualify prospects, book tours, support self-guided visits, and move pre-qualification earlier in the leasing process. When it connects to BoomScreen and BoomReport, applicant information can travel through the funnel instead of disappearing at every software handoff.

The capital will primarily fund product development, more agentic leasing features, new data and fintech products around move-in, and go-to-market hiring. Boom also plans to expand into multifamily and student housing.

Greenstone Biosciences

Greenstone Biosciences just landed a Breakthrough T1D Industry Discovery and Development Partnership grant, and the important word here is not grant. It is derived.

The biotech will build a platform from induced pluripotent stem cells collected from a diverse group of people with type 1 diabetes. Differentiated into 3D pancreatic islet organoids containing insulin-producing beta cells, they give researchers a renewable way to study why the same disease moves differently through different people.

Type 1 diabetes is an autoimmune attack on the pancreas, but clinical and molecular profiles show meaningful variation even among people with similar autoantibody profiles. Human pancreatic tissue is scarce, animal models cannot fully recreate that immune process, and biology keeps reminding researchers that averages are useful until the exception is the patient in front of you.

Greenstone's platform is designed to make that variation visible. The team will expose patient-specific islet organoids to inflammation, metabolic stress, and protective therapies, then examine responses with single-cell RNA sequencing, chromatin-accessibility mapping, proteomics, and metabolomics. The aim is to connect molecular signals with beta-cell vulnerability.

The research is expected to surface actionable pathways, predictive biomarkers, and therapeutic targets that current T1D model systems cannot resolve. Greenstone also plans to bank the iPSC lines as a renewable resource and share summary-level data.

Angelos Oikonomopoulos, PhD, Greenstone's senior scientist and the project's principal investigator, described a scalable multi-omics discovery engine connecting patient biology to therapeutic innovation. Co-founder Joseph C. Wu, MD, PhD, pointed to Greenstone's ability to generate high-quality iPSC lines and differentiated tissues at scale.

The award amount and contract term were not disclosed, and that matters. This is not a venture round, a clinical result, or a treatment approval. It is early research infrastructure aimed at improving the models that decide which ideas deserve the next dollar, experiment, and development cycle.

For operators, the takeaway reaches beyond T1D. Better drug discovery is not only about a smarter algorithm or larger dataset. The underlying model must resemble the humans a therapy is supposed to help. Greenstone Biosciences and Breakthrough T1D are betting that patient-derived biology can narrow that translation gap. Congratulations to the Greenstone team for earning the opportunity. Now the useful question begins: what does human variation reveal when the model stops sanding it away?

Singular Photonics

Singular Photonics just closed an oversubscribed $2.15M funding round, and the interesting part is not that more money found another semiconductor company. It is that the company is moving computation closer to the first moment information exists: when a photon hits a sensor.

ACF Investors led the round, with Wren Capital, Cambridge Angels, Scottish Enterprise, Quantum Exponential, and Old College Capital participating. The stage and valuation were not disclosed.

Singular Photonics says 2026 sales to date have doubled its full-year 2025 sales, while the business is approaching break-even. For a young fabless semiconductor company spun out of the University of Edinburgh, that is cleaner evidence than a valuation headline built to do cardio on LinkedIn.

The company builds SPAD image sensors that detect and time individual photons, then process that information directly on the sensor. Conventional cameras capture light and push the hard work elsewhere. Singular puts photon counting, timing, histogramming, and other computation at the point of detection, helping systems get useful answers faster while moving less data.

That opens a wide field: machine vision, industrial automation, physical AI, scientific discovery, medical imaging, and quantum research. It also explains why the capital is going toward engineering capacity, a broader sensor roadmap, and faster time to market. Customers are asking for products, not another presentation about the future of sensing.

The commercial evidence is stacking. Renishaw integrated Singular's Sirona sensor into a time-resolved Raman spectroscopy module for its inVia microscope. Singular also secured ISO 9001 certification, and NASA published a sole-source notice this month for a prototype camera from the company.

Co-founder and CEO Shahida Imani and co-founder and CTO Dr. Aravind Venugopalan are turning deep research into a commercial semiconductor business. Professor Robert Henderson, whose University of Edinburgh work helped establish the technical foundation, serves as Chief Scientist. Chairman Pete Hutton is now joined by Dr. Dipesh Patel, the former CTO of Arm, on the board.

Semiconductor ambition only matters when brilliant silicon becomes repeatable execution. Singular Photonics has products in market and customers shaping what comes next.

ALSO

ALSO just raised a $150M Series D, led by Prysm Capital with Eclipse and Greenoaks participating. Barely two years after its Rivian spinout, it has assembled $455M in disclosed financing and a valuation reported above $1B.

That number gets attention. The more interesting story is what investors think they are buying.

ALSO is not pitching a nicer bicycle with a software garnish. Chris Yu and RJ Scaringe built the company around a vertically integrated small-vehicle platform: motors, batteries, electronics, software, connected systems, and vehicle architecture designed to move people and goods in form factors smaller than a car.

The first proof point is the TM-B, a Class 3 e-bike starting at $3,500. DreamRide replaces the usual mechanical drivetrain relationship with software-defined pedal-by-wire control. Modular top frames, removable batteries, navigation, and software updates make it closer to a compact EV than a conventional bike.

Then comes the TM-Q. Amazon agreed to buy thousands of customized cargo quads for delivery in the U.S. and Europe. The commercial platform can carry more than 400 pounds while fitting in a bike lane. DoorDash followed with a strategic investment and a multi-year collaboration focused on autonomous delivery.

That is the business case hiding inside the hardware. Dense cities do not need every package, passenger, and grocery run wrapped in 4,000 pounds of steel. They need vehicles matched to the job. ALSO is betting that the space between a bicycle and a van is not a niche. It is an underbuilt transportation layer.

Prysm Capital is leading the Series D after joining ALSO's $200M Series C. Eclipse backed the original $105M spinout, while Greenoaks led the prior round. The same concentrated investor group is funding the move from product reveal into manufacturing and commercial deployment.

The risk is clear. Automotive-grade hardware is expensive, premium pricing narrows the audience, and scaling connected vehicles is harder than producing a beautiful launch video. Autonomy adds capital, regulation, and operational complexity.

But the timing has logic. E-commerce expands the last mile, cities fight curb congestion, and delivery economics punish oversized vehicles. If one stack can support consumer bikes, cargo quads, and autonomous delivery, ALSO gets leverage from reuse instead of starting from zero.

KaliVir Immunotherapeutics

KaliVir Immunotherapeutics just closed a $14M Series A extension, and this is the kind of biotech financing where the capital has a very specific job: keep an ambitious cancer program moving through the clinic without pretending early science has already become a victory lap.

Company K Partners led the round, with participation from SV Investment and Flexus Partners, plus continued support from existing investors including affiliates of Nextrans and Quad Investment Management. KaliVir says the extension brings total capital raised across its Series A financing to $25M. Valuation was not disclosed, which is fine. The operating milestone matters more than the vanity number here.

KaliVir is building oncolytic viral immunotherapies with engineered vaccinia viruses. Its Vaccinia Enhanced Template platform is designed for systemic delivery and expression of therapeutic payloads inside tumors. Lead candidate VET3-TGI expresses IL-12 and a TGF-beta inhibitor, aiming to kill tumor cells while reshaping an immunosuppressive tumor environment.

That candidate is already in STEALTH-001, a recruiting Phase 1/1b study evaluating VET3-TGI through intratumoral injection and intravenous infusion, both alone and with atezolizumab. The trial is early, so nobody gets to fast-forward past safety, dose escalation, or evidence. But the capital is attached to real clinical work: patient enrollment, continued development, and the next milestones for a first-in-human oncology program.

The financing also changes the room where decisions get made. Woo Young Kim of Company K Partners, Kwang Ha Jung of SV Investment, and JP Hong of Flexus Partners joined KaliVir's board. Clinical-stage biotech burns cash and time at industrial scale, so the appointments put capital discipline closer to development discipline.

CEO Helena Chaye and co-founder and Chief Scientific Officer Stephen H. Thorne are steering a company that has moved from platform design into human studies, while Chief Medical Officer James M. Burke and the clinical team carry the less glamorous work that determines whether promising biology survives contact with clinical reality.

The broader signal is not that oncolytic viruses have suddenly become easy. They have not. It is that investors are still willing to fund differentiated platforms when the science has a defined study, a named candidate, and milestones that can be measured without a fog machine.

Ours Privacy

Ours Privacy just raised a $15M Series A to make healthcare marketing measurable without making patient privacy negotiable. Lightbank and Health Velocity Capital led the round, joined by existing investors Rock Health, Lakehouse, TMV, Switch Ventures, Starfire Ventures, and GreyMatter.

That sounds like a compliance story until you see the operating problem underneath it. Healthcare marketers need to know which campaigns work and what deserves another dollar. But the ordinary growth stack was designed for retail clicks, not protected health information. Every handoff creates another place for sensitive data to wander.

Jessica Holton, Adam Putterman, and Tyler Zey learned that problem at Ours Wellness. They protected patient data by stripping out tracking tools, then had to spend with the dashboard lights off. So they built the infrastructure they could not buy and launched Ours Privacy commercially in 2024.

Holton is co-founder and CEO, Putterman is co-founder and Chief Revenue Officer, and Zey is co-founder and Chief Technology Officer. This was not compliance software imagined from a whiteboard. It came from operators who had already paid the price of choosing between visibility and responsibility.

The platform uses server-side connections to control what gets filtered, de-identified, and sent to ad, analytics, EHR, scheduling, and data-warehouse tools. Ours Privacy has expanded beyond a CDP into consent, attribution, testing, session replay, and web scanning.

The company says it serves more than 200 healthcare organizations and processes billions of data points monthly. Those company-reported numbers explain why its new investors pointed to product feedback and healthcare's regulatory constraints.

The useful signal is bigger than privacy software getting another round. Healthcare marketing is becoming infrastructure work. Measurement, consent, activation, and governance cannot keep living in separate tools that pass sensitive information around.

Ours Privacy is betting that the winning system will make compliance the architecture around growth, not a brake applied afterward. The Series A will accelerate product investment and help the company reach more enterprise healthcare organizations. No valuation was disclosed, and the amount of its 2025 financing remains undisclosed, which is where the facts should stay.

Rillet

Rillet just raised a $100M Series C led by ICONIQ at a $1B valuation. Sequoia, Andreessen Horowitz, Sequoia Global Equities, Bain Capital Ventures, Oak HC/FT, Battery Ventures, FirstMark, Scale Venture Partners, and Creandum joined the round. That is a serious cap table for software built around a job most people notice only when the numbers are late.

Rillet is not selling finance teams a shinier dashboard for the same old ledger. Nicolas Kopp and Stelios Modes are building an AI-native ERP where the general ledger, accounting policies, approvals, controls, and AI agents live inside the same governed system.

That distinction matters. Finance automation gets easy in a demo and difficult the moment revenue recognition, multi-entity consolidation, audit evidence, or human approval enters the room. An agent that can draft an answer is useful. An agent that can act against live books, follow policy, preserve a complete audit trail, and know when a person must approve the work is infrastructure.

Rillet says it now serves more than 600 customers and doubled new ARR over the last three months. The company also says customers include public enterprises and fast-growing AI businesses, while larger teams are replacing Oracle Fusion, SAP, Workday, Microsoft Great Plains, and NetSuite.

Those are company-reported numbers, but the financing signal is still difficult to miss. Investors are not only betting that AI will help accountants work faster. They are betting that the system of record itself becomes the place where humans and agents divide the work.

The company has now raised more than $200M, following a $13.5M financing in 2024, a $25M Series A in May 2025, and a $70M Series B just 10 weeks later. Capital does not guarantee category leadership, but this cadence gives Rillet room to expand beyond technology customers into healthcare, biotech, fintech, logistics, and professional services.

The operator lesson is cleaner than the marketing language. AI has more value when it understands the rules, context, permissions, and consequences of the work. That is especially true in finance, where speed without control is merely a faster way to create an expensive meeting.