In partnership with
Good morning, Crowd!
This week, the AI infrastructure wars reached peak absurdity, and I mean that as a compliment.
Anthropic, valued at $900 billion and currently raising $50B more, just signed a deal to rent ALL the compute capacity at Elon Musk's Colossus 1 data center in Memphis. Yes, the same Elon who called Anthropic "evil" on Twitter. Yes, the same Anthropic that's competing directly with his xAI. Elon's response? "I spent the week with the Anthropic team. Nobody set off my evil detector." And just like that, business is business.
Meanwhile, Peter Thiel looked at the power grid crisis, looked at land-based data centers hitting constraints, and decided the solution is obvious: floating orbs in the middle of the ocean powered by waves. He just put $140 million behind it.
And KKR, the private equity giant that usually buys mature companies, announced a $10 billion fund to build AI data centers, power plants, and transmission infrastructure from scratch. Not software. Not chips. Physical infrastructure.
This is what happens when AI scaling hits physical limits. Compute doesn't care about your business model. It needs power, cooling, and chips. And when those run out on land, apparently, we're going to space and the ocean.
Welcome to Saturday. Let's break down the week's biggest confirmed raises. ☕
Jose, Alberto, and the Founderscrowd team.
This week's confirmed raises:
Anthropic rents Elon's data center (300MW, 220K+ GPUs, orbital plans)
Panthalassa raises $140M (Peter Thiel-led, floating ocean data centers)
KKR launches $10B+ Helix Digital (AI infrastructure, not software)
Astranis raises $450M (satellite production, $1.2B total raised)
Quantum Motion lands massive EU round (silicon-based quantum computing)
🎁 Before you keep reading, check this out
Investors are watching this fast growing tech company.
🚨 No, it's not the publicly traded tech giant you might expect… Meet $MODE, the disruptor turning phones into income generators.
📲 Mode’s 32,481% revenue growth ranked them #1 on Deloitte’s list of fastest-growing companies in software. They aim to pioneer "Privatized Universal Basic Income" powered by technology, not government, and their EarnPhone has already helped consumers earn & save $1B+.
Please read the offering circular and related risks at invest.modemobile.com. This is a paid advertisement for Mode Mobile’s Regulation A+ Offering.
Mode Mobile recently received their ticker reservation with Nasdaq ($MODE), indicating an intent to IPO in the next 24 months. An intent to IPO is no guarantee that an actual IPO will occur.
The Deloitte rankings are based on submitted applications and public company database research, with winners selected based on their fiscal-year revenue growth percentage over a three-year period.
🚀 ANTHROPIC RENTS ALL OF ELON'S DATA CENTER (DESPITE BEING "EVIL")
Founderscrowd: Anthropic — the AI company Elon Musk called "evil" in February — just signed a deal to use 100% of the compute capacity at his Colossus 1 data center in Memphis. All 300+ megawatts. All 220,000+ Nvidia GPUs. The entire facility.
And Elon's fine with it. In fact, he spent a week with the Anthropic team and said "nobody set off my evil detector."
The deal:
Capacity: 300+ megawatts (220,000+ Nvidia H100/H200/GB200 GPUs)
Usage: Claude Pro and Claude Max capacity increases
Bonus: Anthropic "expressed interest" in working with SpaceX on orbital AI data centers
Why Elon's okay with this: "SpaceXAI already moved training to Colossus 2"
The context that makes this wild:
Elon merged SpaceX and xAI earlier this year. xAI builds Grok, which competes directly with Anthropic's Claude. Elon has publicly trashed Anthropic multiple times. He literally called them a civilization-hating enemy.
But SpaceX finished building Colossus 2 in January and moved all training there. Colossus 1 was sitting empty. So Elon's renting it to his competitor because... why let perfectly good GPUs sit idle?
Why this matters:
Compute is the new oil. When Anthropic — valued at $900B and raising $50B more — needs to rent an entire data center from a competitor, that tells you compute access is the bottleneck. Not models. Not talent. Compute.
Business beats grudges. Elon called Anthropic evil. Then he became their landlord. This is what happens when everyone needs the same scarce resource (GPUs + power). Ideology loses to economics.
Orbital data centers are real. Anthropic didn't just rent Colossus 1. They "expressed interest" in working with SpaceX on orbital compute. That's not sci-fi. That's a company with $900B valuation planning for space-based AI training.
The Colossus facilities:
Colossus 1 is in a former Electrolux factory in Memphis. Colossus 2 came online in January 2026 with even more capacity. Colossus 3 is under construction in Southaven. xAI is spending $659M on another adjacent building. By 2026, xAI plans to have 1 million GPUs deployed across these facilities.
For context: That's more GPUs than most countries have.
Bottom line: Anthropic paying Elon for compute proves the AI race isn't about who has the best model anymore. It's about who can secure enough GPUs and power to train the next one. When your competitor becomes your landlord because he's the only one with capacity, the game has changed.
🌊 PETER THIEL BETS $140M ON FLOATING OCEAN DATA CENTERS
Founderscrowd: Peter Thiel looked at the AI power crisis — data centers consuming entire power grids, cooling water running out, land-based infrastructure maxed out — and decided the solution is to put data centers in the middle of the ocean and power them with waves.
This is not a thought experiment. This is a $140 million Series B led by Thiel, John Doerr (early Google/Amazon investor), and Marc Benioff. The company is called Panthalassa. It's been building this for 10 years. And they're deploying in 2026.
What Panthalassa actually builds:
Massive floating orbs (85 meters across) that sit in the ocean's most energy-dense wave regions. The orbs:
Generate electricity from wave turbines (consistent 24/7 power)
Use cold seawater for cooling (infinite free cooling)
Run already-trained AI models (inference, not training)
Transmit data via low-Earth-orbit satellites (no cables needed)
Drift with ocean currents (no engines required)
The team includes engineers from SpaceX, Tesla, and NASA. This isn't a startup throwing spaghetti at the wall. This is serious ocean engineering meets AI infrastructure.
The raise:
Amount: $140M Series B
Lead: Peter Thiel
Investors: John Doerr, Marc Benioff (TIME Ventures), Max Levchin (SciFi Ventures), Founders Fund, Lowercarbon Capital, Super Micro Computer
Valuation: Approaching $1 billion
Use: Complete pilot manufacturing facility near Portland, Oregon + deploy Ocean-3 series
Why this isn't crazy:
Wave energy has been a "coming soon" technology for decades. The problem was always: how do you get the power from the ocean to where it's needed? Building transmission infrastructure from offshore to land is insanely expensive.
Panthalassa's insight: Don't transmit the power. Use it right there. AI inference (running already-trained models) can happen anywhere with compute + power + cooling. The ocean has infinite cooling and consistent wave energy. Satellites handle data transmission.
This solves multiple problems:
Land-based power grid constraints (data centers are overwhelming grids)
Cooling water scarcity (seawater is infinite and free)
Permitting delays (ocean deployment skips local politics)
Community impact (no strain on local infrastructure/residents)
Peter Thiel's quote: "The future demands more compute than we can imagine. Extra-terrestrial solutions are no longer science fiction. Panthalassa has opened the ocean frontier."
Why this matters:
When Peter Thiel, John Doerr, and Marc Benioff — investors who've backed PayPal, Google, Amazon, Salesforce — are putting $140M into floating ocean data centers, they're not chasing hype. They're solving a real bottleneck: land-based data centers can't scale fast enough.
For investors: This is infrastructure, not software. If Panthalassa works, it creates a new category (ocean-based compute) with massive moats (ocean engineering + wave energy + satellite networking). If it doesn't, it's a $140M lesson in why you shouldn't fight physics.
Bottom line: AI needs so much compute that we're literally putting data centers in the ocean. This is either the smartest infrastructure bet of the decade or the most expensive proof that some problems can't be solved by throwing money at them. We'll know in 2027 when Ocean-3 deploys.
💰 KKR LAUNCHES $10B AI INFRASTRUCTURE COMPANY (NOT SOFTWARE, PHYSICAL STUFF)
Founderscrowd: KKR — the private equity giant that usually buys mature companies like hospitals and grocery stores — just raised $10+ billion to build AI infrastructure from scratch. Not invest in AI software. Not buy AI chips. Build the physical infrastructure: data centers, power plants, transmission lines, networking.
The company is called Helix Digital Infrastructure. And it's tackling the problem everyone in AI is hitting: compute capacity.
What Helix will actually build:
End-to-end AI infrastructure:
Data centers (purpose-built for AI workloads, not general cloud)
Power generation (natural gas, nuclear, renewables — whatever gets power fast)
Transmission (getting power from plants to data centers)
Connectivity (fiber, networking, interconnects)
This isn't a VC fund betting on startups. This is KKR saying "we'll build the physical backbone ourselves because it doesn't exist."
Why this is significant:
Private equity typically buys existing assets (hospitals, retail chains, infrastructure) and optimizes them. KKR building greenfield AI infrastructure from scratch is a massive signal that:
The bottleneck is physical, not financial. Cloud providers (Amazon, Microsoft, Google) have cash. They're capacity-constrained by power grids, chip supply, and construction timelines. KKR is betting they can build faster.
AI infrastructure is an asset class now. Just like toll roads, airports, and fiber networks became PE darlings in the 2000s, AI data centers + power are becoming standalone investments with predictable cash flows.
Traditional cloud providers can't keep up. When hyperscalers are forecasting record capex ($650B+ in 2026) and still staying capacity-constrained, there's room for a specialized player.
The market opportunity:
Amazon, Microsoft, Google, and Meta are spending $650+ billion on AI infrastructure in 2026 alone. But they're hitting constraints:
Power grids can't deliver enough electricity
Cooling water is scarce in key markets
Chip supply is limited (Nvidia can't make GPUs fast enough)
Construction timelines are 18-24 months (too slow)
Helix is betting it can address the physical-layer bottlenecks by building specialized AI infrastructure that cloud providers can lease. Think: "We build the data center + power plant, you bring the chips and rent capacity."
Why this matters:
When the world's largest private equity firm raises $10 billion to build physical AI infrastructure, that's a signal that AI buildout is moving from software (models, apps) to industrial (power, real estate, construction).
For investors: AI infrastructure is becoming a separate investable category. Not AI models (OpenAI, Anthropic). Not AI chips (Nvidia, AMD). But the boring stuff underneath: power plants, data centers, cooling systems, transmission lines.
Bottom line: KKR doesn't do moonshots. They do boring, cash-flow-generating infrastructure. If they're putting $10B into AI data centers and power plants, it's because they see predictable, long-term demand. The AI boom just went from tech story to industrial buildout.
💎 WHAT PREMIUM MEMBERS ARE READING THIS WEEK
Friday's Premium deal memo:
A Series D enterprise software company doing $180M ARR, 85% gross margins, preparing for 2027 IPO. Already profitable (rare). Raising final private round at $2.8B pre-money. Minimum investment: $100K. Existing investors: Andreessen Horowitz, Sequoia, Salesforce Ventures.
Why it matters: This is one of the last "sleep well at night" enterprise SaaS deals before IPO. Profitable, growing 60% YoY, customers include 40% of Fortune 500. When enterprise software IPOs come back (they will), this exits at $5B+.
This week, Premium members also got:
Fintech Series C ($85M ARR, bank regulators approved, FDIC chair on board)
Defense tech Pentagon contracts ($120M ARR, can't IPO due to security clearances)
Cybersecurity late-stage ($28M ARR, 95% margins, protecting AI companies)
Next week:
A biotech company with FDA breakthrough designation. Phase 2 trials showing 73% efficacy. Raising Series C before Phase 3. High-risk, high-reward. If Phase 3 works, this is a 10x. If it fails, it's zero.
$40/month. Lock in before price increases to $100/month in May.
📡 ASTRANIS RAISES $450M (SATELLITES FOR INTERNET + DEFENSE)
Founderscrowd: Astranis, the company building small geostationary satellites for telecom and internet, just raised $450 million in a Series E led by Snowpoint Ventures and Franklin Templeton. Total funding: over $1.2 billion. This is one of the largest space-tech raises in 2026.
What Astranis builds:
Small, purpose-built satellites for geostationary orbit. Unlike SpaceX's Starlink (thousands of low-Earth-orbit satellites), Astranis puts satellites 22,000 miles up where they stay fixed over one location. This is better for:
Remote internet (Alaska, rural areas, islands)
Government/defense (secure communications)
Telecom backhaul (connecting cell towers in remote areas)
The company's satellites are smaller and cheaper than traditional geostationary satellites, but more reliable than low-Earth-orbit constellations for fixed coverage.
The raise:
Amount: $450M Series E
Lead: Snowpoint Ventures, Franklin Templeton
Total raised: $1.2B+
Use: Ramp up production, scale factory, meet surging demand from commercial + government customers
Why this matters:
Satellite internet used to be a niche market (cruise ships, oil rigs, military). Now it's strategic infrastructure. Governments want secure communications that don't rely on undersea cables (which can be cut). Remote areas want broadband without building fiber. Defense programs want resilient networks.
Astranis sits at the intersection of all three. The $450M raise signals that investors believe satellite communications are becoming critical infrastructure, not just a nice-to-have.
The competitive landscape:
Starlink (SpaceX): 6,000+ low-Earth-orbit satellites, global coverage, latency 20-40ms
Astranis: Geostationary, fixed coverage zones, latency 600ms but more reliable for specific regions
Traditional geostationary (Intelsat, SES): Huge satellites, expensive, slow to deploy
Astranis is faster and cheaper than traditional geostationary players, but more targeted than Starlink. That niche — fixed, reliable coverage for specific regions — is growing fast.
Bottom line: When a satellite company raises $450M, it's not about space exploration. It's about infrastructure. Governments and companies are realizing that internet connectivity is too important to rely solely on ground-based infrastructure. Astranis is building the backup layer.
🔬 QUANTUM MOTION LANDS MASSIVE EUROPEAN QUANTUM ROUND
Founderscrowd: Quantum Motion, a UK-based quantum computing startup, just raised one of the largest VC rounds in European quantum history. The round was led by Mundi Ventures (with EU backing) and DCVC, with participation from British Business Bank, Oxford Science Enterprises, Porsche, and Bosch Ventures.
What makes Quantum Motion different:
Most quantum computers use exotic approaches (superconducting qubits, trapped ions) that require temperatures near absolute zero and custom fabrication. Quantum Motion uses silicon-based quantum computing — the same semiconductor processes that make regular chips.
Why that matters: If you can build quantum computers using existing chip fabs, you bypass the scalability challenges that have plagued the industry for decades.
The round:
Lead: Mundi Ventures (€1B Kembara fund, partly backed by EU institutions), DCVC
Investors: British Business Bank, Oxford Science Enterprises, Porsche, Bosch Ventures
Use: Expand manufacturing, R&D on quantum devices, increase staff in UK + new European labs (recently opened Spain office)
Why EU institutions are involved:
Quantum computing is a strategic priority for Europe. The EU sees quantum as a technology where it can compete with the U.S. and China if it moves fast. Mundi's €1B Kembara fund is partly backed by EU institutions specifically to fund European deep-tech.
This round signals Europe is serious about not ceding quantum leadership.
The market opportunity:
Quantum computers could revolutionize:
Encryption (breaking current encryption, building quantum-safe systems)
Drug discovery (simulating molecular interactions)
Materials science (designing new materials)
Optimization (logistics, supply chains, financial modeling)
The problem: Most quantum computers today are research prototypes. They don't scale. Quantum Motion's bet is that silicon-based qubits will scale because they can use existing chip manufacturing.
Bottom line: When Porsche and Bosch invest in a quantum computing startup, they're not chasing hype. They're betting quantum will solve real industrial problems (materials, optimization, simulation). If Quantum Motion's silicon approach works, Europe could lead the next wave of quantum computing.
🎯 WHAT WE’RE WATCHING
Anthropic's orbital plans:
Does SpaceX actually build orbital AI data centers with Anthropic?
What's the timeline (2027? 2028?)?
How does compute in orbit work (power, cooling, latency)?
Panthalassa's Ocean-3 deployment:
Do the floating orbs survive open ocean conditions?
What's the actual cost per megawatt vs land-based data centers?
Can they scale to hundreds of orbs (as planned)?
KKR's Helix buildout:
Which cloud providers become Helix customers?
How fast can they build vs traditional construction timelines?
Does this create a new model (PE builds infrastructure, hyperscalers lease)?
Astranis production ramp:
Can they scale satellite manufacturing to meet demand?
Do government contracts grow (defense, secure comms)?
How does this compete with Starlink's expansion?
Quantum Motion's silicon approach:
Does silicon-based quantum computing actually scale?
When do we see working quantum computers (not just prototypes)?
Can Europe compete with the U.S./China on quantum?
🚀 THE BOTTOM LINE
Anthropic is renting compute from Elon (despite him calling them evil). Peter Thiel just funded floating ocean data centers. KKR raised $10 billion to build AI power plants. Astranis raised $450M for satellites. And Europe's betting big on silicon-based quantum computing.
The pattern: AI infrastructure is going to extreme places because traditional infrastructure can't scale fast enough.
When you need computing so badly that you're renting from competitors, putting data centers in the ocean, and planning orbital facilities, the bottleneck isn't technology anymore. It's physics.
For investors: The next trillion-dollar opportunity in AI isn't models or apps. It's the boring infrastructure underneath — power, cooling, satellites, quantum chips. The companies solving these hard problems are building moats that software can't replicate.
Bet on the companies building in extreme places. They're solving problems that matter.
See you tomorrow.
Until then, keep looking where others aren't. ☕
Jose, Alberto, and the FC team.
Founderscrowd
P.S. If you're wondering whether floating ocean data centers are genius or insane — join Premium. This week's deep dive breaks down Panthalassa's engineering, the wave energy economics, and whether Peter Thiel just funded the future of compute or the world's most expensive boat experiment. Spoiler: The physics actually works.
DISCLAIMER
Founderscrowd is an independent financial media company. We are not a registered investment adviser, broker-dealer, or FINRA/SEC-regulated entity. Nothing in this newsletter constitutes investment advice, a recommendation, or an offer to buy or sell securities.
Private market and pre-IPO investments carry substantial risk. You may lose some or all of your invested capital. These investments are illiquid and may not be suitable for all investors. Only invest what you can afford to lose entirely.
Deal opportunities are sourced from third-party platforms. Founderscrowd does not act as a broker-dealer or conduct independent due diligence. We may receive compensation for featuring certain opportunities, which creates a conflict of interest.
All Founderscrowd Premium memberships are non-refundable. The value is delivered upon access to information, which cannot be returned once shared.
Past performance is not indicative of future results. Conduct your own research and consult qualified advisers before investing.
© 2026 Founderscrowd. All rights reserved.