• Rillet Reaches Unicorn Status AI-driven accounting platform Rillet joined the unicorn club after securing a $100M Series C led by ICONIQ, highlighting strong VC appetite for back-office automation.

• Lovable Closes $400M Series C AI development tool Lovable raised $400M in a round led by Menlo Ventures to accelerate its next-generation automated workflow tools.

• Physical AI Continues Its Surge Global venture investment in “Physical AI” combining robotics, spatial computing, and hardware continues to accelerate. Funding has reached $47.4B across more than 500 deals so far in 2026, nearly quadrupling late 2025 totals.

• SoftBank Bets Big on Autonomous Machinery SoftBank invested $200M into Gravis Robotics at a $1B valuation to accelerate the deployment of autonomous heavy machinery across construction sites worldwide.

  Thanks for reading SeedWave! Subscribe for free to receive new posts and support my work.

  

AI Infrastructure & Hardware

• Groq Secures $350M AI chipmaker Groq closed a $350M round at a $3.5B valuation to expand capacity for its specialized AI inference hardware.

• Higgsfield AI Skyrockets Generative video platform Higgsfield raised $400M from Goldman Sachs and DST Global, bringing the company’s valuation to $5.4B.

• EliseAI Targets $3.7B Valuation AI property management platform EliseAI is finalizing a $300M round led by Andreessen Horowitz (a16z) and Bessemer.

Logistics & DeepTech Highlights

• ClearJet Raises $25M Air cargo marketplace ClearJet secured $25M in new funding to scale its on-demand freight network.

• Mindgard Secures $30M Series A Cybersecurity startup Mindgard raised $30M led by Album VC to expand its AI defense testing platform.

• Volta Emerges From Stealth AI infrastructure developer Volta officially launched with an initial $5B infrastructure program to build out compute deployments for frontier labs.

1. Base Power and Valar Atomics each raise $1B, the week’s two biggest bets on power

Base Power, an Austin home battery maker, closed a $1 billion Series D at a $13 billion valuation, led by Ribbit Capital, Addition, Valor Equity Partners, and JPMorganChase’s Strategic Investment Group, with a long list of participants including Altimeter, Coatue, Thrive Capital, a16z, and Lightspeed. Its new Base Core battery backs up homes during outages and sells power back to utilities, and the company is already installing about 100 units a day. Total funding to date now tops $2.5 billion.

The same day, Sequoia Capital led a separate $1 billion round for Valar Atomics, a nuclear microreactor startup, valuing it at $6 billion. Valar says the money moves it from demonstrating small reactors to producing them in volume, and it has also lined up a $200 million credit facility. Two very different technologies, same underlying bet: the US power grid needs new capacity fast, and investors want a piece of whoever can add it.

2. Moove raises $250M to build the infrastructure behind self-driving fleets

Moove, founded in Lagos and now headquartered in Abu Dhabi, closed a $250 million Series C co-led by Abu Dhabi’s Mubadala, Toyota’s Woven Capital, and Ion Pacific, with BlackRock and Franklin Templeton also participating, at a valuation of about $2.1 billion. Moove already runs one of the world’s largest human-driven ride-hail fleets (42,000 cars across 13 countries) and partners with Waymo in Phoenix and Miami. Cofounder Ladi Delano describes the new capital as funding the unglamorous part of autonomy: charging hubs, maintenance, data systems, and city-level operations, not just the cars themselves.

3. HappyRobot becomes a unicorn with $150M for enterprise AI agents

US-Spanish startup HappyRobot raised a $150 million Series C led by Prysm Capital and Eurazeo, joined by Andreessen Horowitz, Y Combinator, and Base10, at a $1.2 billion valuation. Its AI agents already automate freight and logistics paperwork for shippers like DHL and Uber, and the new money pushes into insurance, energy, and telecom, industries stacked with the kind of repetitive, multi-step coordination work that’s expensive to set up and hard for a competitor to copy.

4. Sapiom raises $35M to keep AI agents from burning through compute budgets

A smaller round with a good signal: San Francisco’s Sapiom closed a $35 million Series A led by Dragonfly, with Accel and Coinbase Ventures participating, just 11 months after the company was founded and six months after a $15 million seed. Sapiom routes an AI agent’s tasks to whichever model or tool is cheapest and most reliable at each step, aiming to control the runtime costs that stack up once a company is running dozens of agents in production. Anthropic is among its investors. As more startups move agents from demo to production, expect more of this cost control tooling to raise fast follow on rounds.

5. The bigger picture: this week, power was the story

Two of the week’s three largest rounds went to companies solving the same bottleneck from different angles: batteries in homes, reactors in factories. It follows Antora Energy’s $550 million round for industrial thermal batteries just days earlier and Ore Energy’s $43 million raise for iron air storage. Data centers and EVs are pulling more electricity than the grid was built to deliver, and venture capital has decided the winners will be whoever removes that constraint, however unglamorous the hardware looks on a pitch deck.

The takeaway: the physical constraints, power, compute, hardware, are where the money’s going right now. Teams that can point to a specific bottleneck they’re removing are raising faster than teams pitching a vision.

Startup capital didn’t slow down this week, it just got more specific. Investors are chasing companies that own a real workflow (tax, compliance, security, energy, sales) instead of ones selling a vibe. Here are the five stories shaping the week, and what they signal for where the next wave of startups is headed.

1. Atoms raises $1.7B to build “physical AI,” the week’s biggest round by far

Travis Kalanick’s newest venture, Atoms, closed a $1.7 billion round led by Andreessen Horowitz, with Ben Horowitz taking a board seat. New investors include Bain Capital, Fifth Wall, Chemistry, A*, K5 Global, Abstract, SV Angel, and Alpha Square Group, and Uber itself also joined the round, reconnecting Kalanick with the company that pushed him out as CEO in 2017. “Physical AI” (AI systems that operate in the real world rather than just on a screen) is quickly becoming the label VCs reach for to describe the next act after the LLM boom. It’s part of a record year for robotics funding, which has already hit $55.8 billion globally.

2. Meshy AI hits a $1.5B valuation on $400M Series B

Meshy AI, a startup building foundation models for AI-generated 3D content, raised nearly $400 million in a round co-led by IDG Capital, Matrix Partners China, and Monolith, at a $1.5 billion valuation. It’s the largest round to date in AI 3D generation, a category that’s been quietly heating up as gaming, VFX, and industrial design teams look for AI shortcuts to 3D asset creation. The company says its annual recurring revenue is growing about 12x year over year, with over 12 million registered users.

3. Sila raises $300M to scale US battery anode production

Battery materials company Sila closed a $300 million round led by Atreides Management and Sutter Hill Ventures, with 8VC, Bessemer, Matrix Partners, and T. Rowe Price also participating. The money will expand its Moses Lake, Washington plant, adding enough silicon-carbon anode material for more than 100,000 EVs. Battery materials sit at the intersection of AI infrastructure (data centers need power) and the EV supply chain, two theses VCs are betting will only get more capital-hungry.

4. Centralize emerges from stealth with a $15M Series A

Smaller than the megadeals above, but a good signal in its own right: enterprise sales platform Centralize came out of stealth with a $15 million Series A led by NEA, joined by Salesforce Ventures, Y Combinator, and angel investors including Slack co-founder Stewart Butterfield. Combined with an earlier seed round, the company has raised $19 million since 2023 to build what its founders call a “deal GPS” for enterprise sales teams, already used by Sierra, ElevenLabs, Cognition, and Brex.

5. The bigger picture: money is concentrating, not spreading out

Crunchbase’s weekly roundup calls it a “varied week for large deals,” but the pattern underneath is consistent. Capital is flowing to AI infrastructure, fintech, biotech, cybersecurity, and defense: sectors where startups control something customers already need and budget for. Separately, the US government gave OpenAI the green light to roll out its most advanced model globally after a staggered preview, one more sign of how much weight “government approval” now carries in the AI race. That demand is already straining memory chip supply: Gartner projects prices could climb roughly 130% by the end of 2026, pushing PC prices up 17% and smartphone prices up 13%.

The takeaway: if you’re building or backing something right now, “polished story” isn’t the unlock anymore. The rounds getting done are going to teams solving an expensive, specific problem, fast.

The traditional “College for All” pipeline is broken. Right now, over half of recent graduates are working in jobs that don't require bachelor's degrees, and a decade into their careers, 45% of them remain underemployed. Meanwhile, high-paying, fulfilling, and AI-resistant "gray collar" jobs are sitting empty. Don Smith, co-founder of Tanager Holdings and the builder behind Gray Collar Collective, joins us to explore how he is building the infrastructure to route students into these high-growth careers and why "Gray Collar" is set to be the defining term of 2026.

Subscribe now

We discuss:

• The “broken pipeline” and why the traditional education system is failing the modern student

• Defining Gray Collar and why these roles are the ultimate hedge against AI disruption

• Don’s non-linear journey from the Air Force Academy and law to fintech and venture studios

• The “Elevator Pitch” and real-world stories of engineers finding six-figure success in the trades

• Building the infrastructure for self-discovery, training, and placement in the skilled workforce

• The Jacksonville innovation ecosystem and the importance of hubs like the Jax Innovation Hub

• Hard-earned intuition and why Don’s top advice for founders is to “validate aggressively” and let the market “kick your baby”

Featured Resource Stack:

• Gray Collar Collective: www.graycollar.com

• Jacksonville Innovation Hub: https://www.jaxinnovationhub.com/

• Perspectiv (Meagan Skerchock): https://www.getperspectiv.com/

• Champion Now (Terry Iverson): https://championnow.org/

• Cardinal Manufacturing ( Craig Cegielski): https://www.cardinalmanufacturing.org/

Big influencers are charging over $1,000 per reel while delivering engagement rates below 3%. Meanwhile, 76% of everyday social media users have fewer than 10,000 followers, and 88% believe they deserve a reward for authentically promoting the brands they love. Swayze Co-CEO & Co-Founder Phi Long Bui joins us to share how his platform is bridging this gap by turning everyday restaurant-goers into powerful nano-influencers, completely flipping the script on local digital marketing.

Subscribe now

We discuss:

1. How a background in strategy consulting led to the founding of Swayze

2. Why traditional influencer marketing is broken for small businesses

3. What a “nano-influencer” is and why they drive better ROI than celebrities

4. The stark differences between scaling a startup in Poland versus New York City

5. How partnering with intermediaries will accelerate their B2B growth

6. The future of social rewards and why retail is the next big frontier

7. Why his ultimate advice to young professionals is to just start building

If you’re running a waste business, your biggest headache isn’t the trash, it’s the office.

Lately, I’ve been obsessed with the invisible infrastructure that keeps our world turning. After exploring robots in data centers with Boost Robotics and computer vision for construction via Scout Out AI, it was only a matter of time before I looked at the fleet.

The “Open-Air Factory”

CEO Mark Hoadley once described the reality of this business as “running a factory with no roof and no visibility into what’s happening in that factory” (TechCrunch via aVenture). In a traditional factory, you have sensors and quality control. In waste hauling, your factory floor is the entire city, and for decades, that floor has been a total black box.

Hauler Hero is building the Central Operating System that finally puts a roof over that factory. They’ve replaced fragmented legacy tools with a single, cloud-based hub. But their true Wall-E moment comes from their AI agents. Hero Vision uses truck-mounted cameras to see the world in real-time, automatically documenting every pickup. If a bin is overloaded, the AI tags it and the customer is billed, moving the industry away from vibes-based pricing toward data-driven accuracy.

The Infrastructure Supercycle

To understand why Hauler Hero is a heavy hitter, you have to look at the current Infrastructure Supercycle. We are seeing a shift toward hard-tech solutions that solve boring, massive, trillion-dollar problems. Waste management is a $145B industry in the U.S., yet it’s one of the last to be digitized.

By saving office employees an average of 14 hours a week, Hauler Hero isn’t just a convenience, it’s an operational necessity. But the real Aha moment for owners is the found revenue. Early partners using the system have reported 19% faster cash collection and a 15% reduction in customer service calls. By automating not-outs and overloaded bin detection, they are moving away from vibes-based pricing and recapturing thousands in fees that used to simply leak away. As the market sees massive consolidation, independent haulers are hungry for this high-tech armor to help them stay competitive.

Physics vs. Algorithm

When we talk about AI in the dumpster, we have to ask the hard questions about how this survives the real world. Computer vision is notoriously finicky when it leaves the clean environment of a server room.

In the waste industry, clean data is an oxymoron. We have to consider the Dirty Lens Problem, where the AI must differentiate between a truly overloaded bin and a camera lens covered in North Florida mud or morning frost. The architectural choice between edge or cloud will quietly determine whether this becomes infrastructure or remains a pilot project. A waste truck is not a server room. It’s vibration, glare, mud, and spotty signal. The AI has to survive physics.

Finally, there is the vulnerability to vibes. While the system aims to end subjective pricing, AI models still struggle with depth perception and visual clutter. A system needs to be incredibly sophisticated to tell the difference between a bin that is legitimately over-full and one where a single piece of cardboard is just sticking out at a weird, misleading angle.

The verdict

Wall-E didn’t just move trash, he cleared the path for a new beginning. That’s exactly what Hauler Hero is doing for the waste industry. They are taking the unsexy work of estimation and routing and turning it into a high-tech operation. In the world of infrastructure tech, they aren’t just riding the wave. They’re clearing the path.

⚡ The Wave Report

• Founders: Mark Hoadley and Ben Sikma

• The Scale: 35M pickups annually and over $300M in annualized GMV

• The Mission: Building world-class technology with easy-to-use design for the "hardworking heroes" of the waste and recycling industry.

• Find them: haulerhero.com

Right now, if you want to price out a kitchen remodel or a full home renovation, your Day Zero doesn’t happen on the job site. It happens at 9:00 PM at your kitchen table. You’re squinting at a set of PDFs, wrestling with a tape measure, and trying to guess if you need 40 or 50 sheets of drywall.

It’s called a takeoff, the construction version of a master shopping list, but it feels more like a prison sentence. It takes hours. It’s prone to error. And if you’re off by 10%, your profit margin for the entire month just evaporated.

In a world where we can generate an entire video from a text prompt, why are we still doing construction math like it’s 1950?

Enter Scout Out

I’ve been digging into the “unsexy” side of AI, the kind that actually “touches” dirt and wood, and Scout Out (Y Combinator W26) caught my eye for one reason: They want to kill the manual estimate.

The way this actually works is surprisingly human. You don’t have to be a tech wizard or spend an hour uploading blueprints.

Instead of fighting with a scanner, you just talk to the app like you’re talking to a partner. You tell it what’s happening: “Hey, we’re looking at a 200 sq. ft. master bath gut-reno. We’re doing a walk-in shower, double vanity, and we’re going with marble tile.”

That’s it.

The AI isn’t just processing data. It’s actually listening to the logic of the build. It knows that a walk-in shower isn’t just a shower. It knows exactly how much thin-set and grout you’ll need to make that marble stick. It understands the labor hours that go into a double vanity.

In about 60 seconds, it hands you back a professional, line-itemed proposal. Think about that for a second. While you’re still standing in the client’s driveway, before you’ve even put the truck in reverse, you’ve handed them a quote. You aren’t going home to work the night shift at your kitchen table anymore. You’re already done.

But is the math right?…

This is the part that usually makes people lean back. If an AI hallucinates the price of 500 square feet of marble, someone is losing a lot of money.

But Scout Out is built as a Co-pilot, not an Autopilot. The AI gives you the baseline in 60 seconds, but the contractor still has the final word. You can go in and adjust your markups or tweak the labor rates to match exactly what you pay your crew. It’s not about the AI being 100% perfect on day one; it’s about it being fast enough to get a professional quote to a client before they even call the next guy.

Why is this a game changer?

The “Aha!” moment here is speed. In construction, the first person to get a professional quote to the homeowner usually wins the job. Most contractors lose work because they “get to the math” three days too late. Scout Out turns the estimate into a conversation you have on your phone while the lead is still hot.

Additionally, Scout Out is currently on the ground in Los Angeles, working with contractors on the high-stakes rebuilds following the Palisades and Eaton fires. In an environment where families have lost everything, speed is everything. If you can’t get a quote out, you can’t get insurance money, and you can’t start building.

By automating the boring work of estimation, they are clearing the path so the builders can actually get back to work.

The verdict

The more I explore these startups, the more I realize the Intelligence Age is finally hitting the pavement. Scout Out isn’t just measuring a room. They’re fixing the foundation of how we build.

They aren’t just riding the wave. They’re clearing the path.

⚡ The Wave Report: Scout Out

• The Problem: Contractors spend 14+ hours a week on “unpaid” office work (estimating).

• The Solution: Text-to-Estimate AI that builds quotes in <60 seconds.

• The Edge: A mobile-first CRM that lets you bid, sign, and manage leads from the truck.

• Check them out: scoutout.ai

But as the world looks at the screen, nobody is looking at the floor.

Data centers are the massive, overheating warehouses that make your ChatGPT prompts possible. They are the beating hearts of 2026. Right now, a quiet crisis is brewing inside them. It isn’t a software bug or a bad algorithm. It is a physical limit.

The AGI Bottleneck

As we scale toward AGI (Artificial General Intelligence, AI that can perform any intellectual task a human can), the sheer volume of hardware required is exploding. We are moving from clusters of 10,000 GPUs to campuses with millions of them.

These facilities are loud, they are hot, and they are miles long. When a server rack fails or a cooling pipe leaks, a human has to walk those miles in 100-degree aisles to find the fault. In a world where compute is the most valuable commodity on earth, we are running out of humans who can work that fast.

This is where the “Day Zero” energy lives. While the mega-caps are throwing money at models, a two-person team in Boston is building the mechanics that will keep the lights on.

Who is behind Boost Robotics?

I’ve been tracking the Physical AI wave in Boston, and the team at Boost Robotics caught my eye. Hans Kumar and Hardik Singh are a two-person squad with a background that suggests they know exactly how hard this problem is to solve.

According to their Y Combinator founder profile, their bond goes way back to 2015, when they met as engineering freshmen at CMU. They actually spent years together on a competitive Bhangra dance team before they ever built a robot together.

Hans was a Staff Software Engineer at Boston Dynamics (the world leader in mobile robotics), where he worked on productizing the Spot inspection robot. Hardik was the robotics lead at an agtech startup where he successfully pivoted a product to market in just one month.

The Market Landscape

To understand why a tiny team like Boost Robotics is a heavy hitter, you have to look at the 2026 market. We are currently in an Infrastructure Supercycle, which is a multi year investment phase where companies spend more on “the foundation” - buildings, power, and hardware - than on the software itself. Global spending on AI is forecast to reach a staggering $2.52 trillion this year. Within that, hyperscalers like Microsoft and Google are expected to spend over $500 billion specifically on infrastructure.

The industry has moved from the Training era, where the focus was on building models, to the Inference era, where those models are actually deployed for users. This shift requires “always on” reliability. The biggest threat to the boom isn’t a lack of code, it’s now a lack of uptime.

The Verdict

The East Coast is becoming a powerhouse for Physical AI. We are seeing a shift away from “another SaaS app” toward hard-tech solutions that solve boring, massive, trillion-dollar problems. Keep an eye on Boost Robotics. They are moving fast to ensure the physical world can keep pace with the digital one. They are not just riding the wave. They are building the surfboard.

⚡ The Wave Report

• Team: 2 founders (Hans Kumar & Hardik Singh)

• Status: Active (Y Combinator Spring 2025 Batch)

• The Mission: Build autonomous mobile robots that perform manual tasks in data centers

• Find them: boostrobotics.ai

Every great journey starts with a single spark. Before the Series A rocket ship, and before the Seed round’s powerful engine, there’s the Pre-Seed stage.

In the ecosystem of innovation, this is arguably the most exhilarating and foundational period in a startup’s life. It’s the “garage phase”, where your idea first learns to drive.

The Napkin Sketch & The Garage

Imagine you have a brilliant idea for a revolutionary vehicle. You’ve got it all sketched out on a napkin. You believe it can change how people move, but right now, it’s just a sketch. You need capital to buy parts, hire a skilled mechanic (your co-founder!!), and build a working prototype in your garage.

Pre-seed funding is the fuel for that prototype. It isn’t about market dominance yet, it’s about proving your core hypothesis. It transforms the “what if” into a functional Minimum Viable Product (MVP).

What Matters Most at Day Zero?

At this stage, investors aren’t looking at your revenue spreadsheets, they’re looking at your trajectory. Specifically, they are betting on three things:

1. The Founders: Can this team actually build the vehicle? They look for grit, obsession, and technical “founder-market fit.”

2. The Problem: Is this a minor inconvenience or a massive pain point? The bigger the problem, the more room your vehicle has to run.

3. The Prototype (MVP): Do you have a plan to build a version that actually moves? It doesn’t need a stereo system yet, it just needs to get from A to B.

Where Does the “Fuel” Come From?

Unlike later rounds led by massive VC firms, pre-seed capital usually feels more personal:

• Friends & Family: Your day-one believers.

• Angel Investors: Experienced individuals taking early-stage bets.

• Specialized Pre-Seed Funds: Micro-VCs like First Round Capital or Pear VC that specialize in this “niche” chaos.

From Garage to Gear Shift

The ultimate goal of the pre-seed stage is to gather enough evidence to prove the car is worth driving. Once you’ve clocked those first few miles and gathered user feedback, you’re ready for the Seed Round, where you get the fuel to finally hit the highway.

The Hook

I first felt it in a crowded hall during a grad school pitch competition. I was standing there, defending an idea, and I realized that I wasn’t just interested in the business model—I was addicted to the energy of the room. It was that specific, high-stakes electricity of an idea meeting reality for the first time. I haven’t been able to shake that feeling since, and it’s exactly why I’m launching this project.

The “Why”

I’m starting this Substack and Podcast because I’m tired of “hindsight” content. Most startup advice is written by people looking back from the finish line, which makes everything look like a clear, logical path. I want to talk about the build while it’s still messy.

I care about these topics because:

• The Rush of the Messy Middle: The real thrill isn't in the $100M exit; it’s in the high-stakes adrenaline of the week a founder almost missed payroll, or the lightning-bolt moment they realized their product had to change.

• Building is a High-Stakes Craft: I view startup growth as an art form. From the way a founder tells their story to the way they hire their first employee, there is a craft to it that deserves to be highlighted.

• The Energy of Innovation: Ever since that pitch competition, I’ve been fascinated by the “Day Zero” hustle. I want to spend my time with the people who are brave enough to start something from nothing.

What you’ll find here

This is a hobby-driven project, which means I can be honest. You can expect:

1. The Growth Playbook: Real-time analysis of how early-stage companies are finding traction in 2026.

2. Founder Conversations (The Podcast): I’m sitting down with founders to talk about the stuff they usually edit out of their LinkedIn updates—the doubts, the near-misses, and the gritty details of scaling.

3. Zero-to-One Deep Dives: Breaking down the specific frameworks that help a startup go from an idea to a growing entity.

The Build Continues

I’m here to learn as much as I am to share. If you’re a founder, an early-hire, or just someone who loves the puzzle of a new venture, I’m glad you’re here. Let’s get into the build together.

Subscribe now