In partnership with
Saturday, April 25, 2026 | Private Markets Intelligence
Good morning. A nuclear startup went public at $12B before building a single reactor. Google and Amazon are literally buying the same AI company for a combined $65B+. And Palantir got paid $130M to help the IRS hunt your Bitcoin.
This is what happens when AI needs more power than entire countries—and governments need to know where your money went.
Jose, Founderscrowd
In today's rundown:
X-Energy IPOs at $12B (zero reactors built, AI power play)
Google drops $40B on Anthropic (after Amazon's $25B two days earlier)
ComfyUI hits $500M valuation (fixing AI's "six-finger problem")
Palantir's IRS deal exposed ($130M to track crypto, taxes, everything)
Nuclear + AI infrastructure = the only game that matters
Deloitte: Robot “Adoption is Accelerating Exponentially”
Robots are going from niche to mainstream, per Deloitte. They say it’s especially true in places where “physical AI solves real problems.” Take the $1 trillion fast-food market, where brands turn to robots to alleviate 144% labor turnover.
At the forefront of this push is Miso Robotics, which is celebrating a decade of robotic innovation by preparing for its most impactful year yet.
Miso’s Flippy Fry Station AI robot has already been adopted by major brands like White Castle, frying 5M+ baskets of food to date. That earned strategic investment from industry powerhouse Ecolab and an unique collaboration with NVIDIA.
Now, after acquiring Zignyl, the powerful restaurant-operations tool, Miso adds powerhouse operators like Cinnabon, Jamba, and Jersey Mike’s under their umbrella.
Next up? Miso’s scaling across a $4B/year revenue opportunity. Join 39,000+ people as an early-stage Miso investor before they reach 100,000+ target locations.
This is a paid advertisement for Miso Robotics’ Regulation A offering. Please read the offering circular at invest.misorobotics.com.
⚛️ NUCLEAR STARTUP IPOS AT $12B WITH ZERO REACTORS BUILT
Founderscrowd: X-Energy went public at $12B (up 27% day one) with zero reactors built and zero licenses. Amazon's backing the bet: AI needs nuclear power or it stops growing.
Why it matters:
AI's hitting a wall: data centers are maxing out power grids and states are banning new ones. X-Energy's pitch is small nuclear reactors built next to data centers for dedicated power.
The catch? They haven't built anything yet. No reactor. No approval. No customers. Just Amazon writing checks because the alternative is worse—AI companies fighting over limited electricity while growth stalls.
Nuclear's the only tech that's carbon-free, always-on, and big enough for AI's scale. Google and Microsoft are making similar bets. If X-Energy delivers one reactor by 2030, Amazon orders 50 more. If they miss, $12B evaporates.
Bottom line: Pure infrastructure bet—AI needs nuclear or it can't scale. Speculative? Yes. But whoever solves AI's power problem wins billions.
💰 GOOGLE DROPS $40B ON ANTHROPIC (AFTER AMAZON'S $25B)
Founderscrowds: Google committed $40B to Anthropic on Friday. Amazon committed $25B on Monday. Total this week: $65B for the same AI company. Both deals lock Anthropic into their clouds.
Why it matters:
This isn't about owning Anthropic—it's about proving their chips work. Amazon needs Anthropic to validate Trainium chips. Google needs Anthropic to validate TPUs. Microsoft already has OpenAI locked to Azure running on Nvidia.
The playbook is simple: "Use our chips exclusively, we'll give you billions in cash + compute." Anthropic gets $65B to scale. Amazon and Google get proof their custom chips can replace Nvidia.
If it works, they save billions and undercut Microsoft on AI cloud pricing forever. If Trainium and TPUs fail, they're stuck paying Nvidia premiums while Microsoft wins.
For scale: Anthropic's locking up 10+ gigawatts of compute—enough to power 7.5 million homes. That's AI infrastructure reality now.
Bottom line: $65B in one week proves compute is the real battle. Custom chips (Trainium, TPUs) vs. Nvidia decides who wins AI cloud wars. Amazon and Google can't afford to lose.
WANT IN ON DEALS LIKE THESE? (LAST WEEKEND AT CURRENT PRICE)
Most of these opportunities disappear before they hit the news.
Founderscrowd Premium members get first look at exclusive pre-IPO investments, the kind of deals institutional investors see before they go mainstream. We're talking early-stage access to companies that could be the next exits in satellite, EVs, fintech, and AI.
Premium includes:
✅ Weekly deal memos on live investment opportunities
✅ Private markets intel before it's public
✅ Direct access to carry on winning deals
✅ Community of 1,000+ active private markets investors
The difference? When OpenAI acquires a company, Premium members already knew the founders. When Amazon drops $11.57B, Premium members already owned the company. When Tesla hits 10M subscribers, Premium members will have already ridden the private equity wave.
Become a Premium member →
Lock in $40/month before it goes to $100+ (last week)
🎨 COMFYUI HITS $500M VALUATION FIXING AI'S "SIX-FINGER PROBLEM."
Founderscrowds: ComfyUI raised $30M at $500M valuation. Gives creators precise control over AI images and video. Started in 2023 fixing Midjourney's hand problem. Now has 4M users and made the first AI Super Bowl ad.
Why it matters:
AI generators get you 80% there, then you're stuck. Change one thing, the whole image regenerates differently. CEO calls it "slot machine" prompting.
ComfyUI fixes this with node-based control—isolate the part you want to change (sky, lighting, face), adjust only that, everything else stays locked. It's Photoshop-level precision for AI.
Professionals need this. Ad agencies can't gamble on lottery tickets. Studios can't regenerate entire scenes for lighting tweaks. "ComfyUI artist" is now a real job title. Silverside AI used it for SVEDKA's 2026 Super Bowl ad—first AI-generated Super Bowl commercial.
Started open-source (still is). Now sells hosted workflows, APIs, and enterprise licenses.
Bottom line: $500M says control beats convenience. Prompts work for one-offs. Pros need repeatable workflows. As AI gets better, precision demand grows—not shrinks.
🕵️ PALANTIR'S IRS DEAL EXPOSED: $130M TO TRACK YOUR CRYPTO
Founderscrowds: Palantir's been helping the IRS Criminal Investigation office since 2018, paid $130M. Their software connects your tax returns, bank accounts, crypto wallets, and dark web activity into one platform. Finds "millions of records with thousands of links" instantly.
Why it matters:
Palantir doesn't just read your taxes—it connects everything. Example: You file in California, wire $50K to Delaware, that person buys crypto on Coinbase, sends it to a dark web wallet, cashes out in Hong Kong. Palantir's software maps this across databases in seconds.
Started in 2018, expanded under Biden, now in Trump's hands. The IRS Criminal Investigation office has "pivoted toward investigating left-leaning groups" (WSJ Oct 2025). Privacy groups call it "massive-scale data mining with minimal transparency."
IRS says it's for catching money launderers and dark web dealers. Former officials worry about "lock-in"—agencies can't leave Palantir without rebuilding everything. 13 ex-Palantir employees warned "guardrails to prevent abuse have been violated."
Bottom line: Palantir's model is sell indispensable software, agencies can't leave, charge whatever you want. $130M IRS is one contract—they have similar deals across DOD, CIA, ICE, DOGE. Privacy controversy hasn't slowed growth—it's increased demand.
THE BIG PICTURE
What this week reveals:
1. AI can't scale without nuclear power. X-Energy went public at $12B with zero reactors because AI companies have run out of grid capacity. Data centers need dedicated nuclear or they hit a ceiling. Amazon, Google, Microsoft are all betting on SMRs.
2. Compute is the new oil. Google and Amazon committed $65B to Anthropic in one week—not because they love Claude, but because they need proof their custom chips (Trainium, TPUs) can replace Nvidia. Whoever controls cheap compute wins the AI cloud wars.
3. Control > convenience in AI tools. ComfyUI hit $500M valuation solving a simple problem: AI generators work 80% of the time, but professionals need 100%. The market for "AI that actually works in production" is bigger than the market for "AI that's easy to use."
4. Governments are buying surveillance infrastructure. Palantir's $130M IRS contract reveals how far data aggregation has gone. Your taxes, bank accounts, crypto wallets, and dark web activity (if any) are all in one searchable platform. The playbook: sell indispensable software, charge whatever you want, repeat across every agency.
5. Infrastructure wars define the decade. Nuclear reactors, custom chips, compute lockups, data platforms—this is where the money is. Apps commoditize. Infrastructure captures value.
WHAT TO WATCH
X-Energy execution:
Can they actually build a reactor by 2030?
Does regulatory approval happen or drag out for a decade?
Stock volatility if timelines slip (likely)
Anthropic's compute crunch:
Do Amazon's Trainium chips and Google's TPUs actually work at scale?
Can Anthropic handle 10+ gigawatts of capacity without infrastructure failures?
OpenAI's next move (they just mocked Anthropic's capacity problems)
ComfyUI enterprise adoption:
Does $500M valuation justify itself with enterprise revenue growth?
Can they monetize Comfy Cloud/API beyond open-source community?
Competitor response (Figma acquired Weavy last year)
Palantir's government expansion:
More contracts exposed via American Oversight lawsuit?
DOGE project details (what exactly is Palantir building for Trump?)
Stock reaction to privacy controversy (historically: none, stock rises anyway)
That's your Saturday breakdown.
A nuclear startup IPOed at $12B before building anything. Two tech giants committed $65B to the same AI company in 72 hours. A creative tool hit $500M valuation fixing AI's finger problem. And Palantir got paid $130M to connect every dot in your financial life.
The pattern: whoever controls the infrastructure underneath—power, chips, compute, data—wins. Apps are commodities. Infrastructure is the moat.
⭐⭐⭐⭐⭐ This is the signal
⭐⭐⭐ Average
⭐ Needs work
See you Sunday,
Jose ☕
Founderscrowd
P.S. Sunday we're breaking down why AI companies are spending more on infrastructure than some countries spend on defense. The compute wars are real, and they're about to get expensive.