Ex-OpenAI VP and DeepMind materials chief just raised the largest seed round of 2025.

Periodic Labs closed $300M seed to build AI scientists that automate materials discovery. Founded by Liam Fedus (ChatGPT co-creator, ex-OpenAI VP) and Ekin Dogus Cubuk (led DeepMind's GNoME project that discovered 2M+ new crystals).

• $300M seed led by a16z, DST, Nvidia, Accel
• Backers: Jeff Bezos, Eric Schmidt, Jeff Dean
• Pre-money valuation: $1B
• Goal: Room-temperature superconductors
• Team: 25+ researchers from Meta, OpenAI, Samsung

How it happened:
• Cubuk built GNoME at DeepMind → discovered 2M crystals
• Fedus co-created ChatGPT → left OpenAI as VP Research
• March 2025: Co-founded Periodic Labs
• Recruited 25+ top AI/materials researchers
• Pitched autonomous labs + fresh training data angle
• Closed $300M from tech elite in 6 months

Takeaway: Elite AI talent + hard science focus = mega rounds.

Source: TechCrunch

Brewed by Startup Beaker.

❌Most founders waste months building products based on fake validation from people who tell them “yeah that’s a cool idea” but never pay.

✅The Mom Test teaches you how to ask questions that cut through politeness and uncover what users actually need and will spend money on

More available on:
https://t.me/+4MYt0sh0PUBmMGRi

DualEntry raised $90M Series A from Lightspeed, Khosla Ventures, and GV.

The AI-native ERP processed $100B in transactions and serves companies from $5M to NYSE-listed.

Founder playbook:
• Scaled previous company to $100M ARR
• Failed ERP implementation = product insight
• Built AI-first vs retrofitting legacy systems
• 24-hour migration vs 18+ month industry standard
• 90% manual task automation
• 13,000+ integrations from day one

Key metrics: $100B processed, $100M+ funding in 18 months.

The wedge: Experience the pain, build the solution.

When founders who scaled $100M businesses build ERP software, they know exactly what other founders need.

https://www.dualentry.com/funding-announcement

Ex-OpenAI VP Liam Fedus and DeepMind materials lead Ekin Dogus Cubuk just raised a record-breaking $300M seed for Periodic Labs.

What they're building: Autonomous AI labs that physically synthesize materials and run experiments. Not simulations—real robots mixing compounds, heating materials, discovering superconductors.

Why it matters: AI models trained on internet text hit a wall in physical sciences. Periodic closes the loop between hypothesis and reality using nature as the RL environment.

How it happened:
• Realized frontier models fail at physics/chemistry
• Built team of physicists + top ML researchers
• Landed customers in space, defense, semiconductors
• Raised from a16z, NVIDIA, Bezos, Schmidt

The play: $15T industries untouched by text-only AI.

https://techcrunch.com/2025/09/30/former-openai-and-deepmind-researchers-raise-whopping-300m-seed-to-automate-science/

Brewed by Startup Beaker.

Female founder breaks records with $64M seed for AI mathematician

Carina Hong closed the largest seed round of 2025 — $64M at $300M valuation for Axiom Math, an AI that solves complex math problems and generates new mathematical knowledge.

Key differentiators:
• AI produces step-by-step proofs verified by Lean and Coq
• Transforms English math from textbooks into executable code
• Team includes Meta FAIR lab veterans

Founder playbook:
• Assembled world-class team before fundraising
• Built working proofs in specialized domains
• Demonstrated clear commercial applications (finance, chip design)
• Positioned as foundational AI infrastructure
• Attracted tier-1 VCs: B Capital, Greycroft, Madrona

Takeaway: Deep tech needs deep expertise first, funding second.

Source: Tech Funding News

#startups #funding

Brewed by Startup Beaker.

Solo GP Sheena Jindal just closed a $32M inaugural fund targeting MIT technical founders.

Sugar Free Capital focuses on AI-native infrastructure with a systems engineering mindset. Jindal, former Bessemer/Comcast Ventures investor, targets 15 companies with $1M-$5M checks.

• $32M fund from family offices (Nvidia, Citadel execs)
• 4 companies already backed (defense, gaming, workflow automation)
• Quarterly themes: physical AI, data center optimization, AI agents
• 4-5 deals per year target
• MIT alumni gap: few early-stage investors despite lucrative exits

How it happened:
• Identified "too sugary" 2021 valuations at CV
• Built thesis around "age of intelligence" + technical founders
• Leveraged MIT network gap vs Harvard/Stanford
• Secured LPs who valued MIT access + clear thesis
• Launched with concentrated portfolio strategy

Takeaway: Niche focus + network gaps = fund differentiation.

Source: TechCrunch

Brewed by Startup Beaker.

Driver cuts semiconductor documentation from weeks to hours with AI.

What it is: AI platform that transforms thousands-page chip manuals into user-specific documentation. Targets both chipset suppliers and engineers who implement them.

Key metrics:
• $8M seed led by GV
• Promises 50% faster codebase understanding
• Saves weeks of employee documentation time
• Language-agnostic platform

How it happened:
• CEO Adam Tilton (ex-Nike, sold previous startups) faced documentation pain repeatedly
• Used ChatGPT to extract APIs from PDFs manually
• Co-founder suggested building platform at scale
• Targeted both suppliers and implementers
• Secured GV backing with proven founder track record

Why it matters: Technical documentation is a massive time sink. Driver shows how AI can tackle unsexy but critical B2B workflows.

Source: TechCrunch

Brewed by Startup Beaker.

Zego alumni raised £2.6M pre-seed for AI that picks your tech stack.

Dragonfly launched conversational AI that recommends software from 250,000+ tools in seconds. Founded by Sean King and Sven Sabas in December 2024.

Key details:
• £2.6M led by Episode 1
• Angels: QuantumBlack founder, Bolt CEO
• 12-person team from successful exits
• Enterprise rollout planned late 2025

How they built it:
• Identified pain: too many software options
• Built world's largest software catalog
• Developed conversational AI interface
• Secured experienced team and early customers

Handles simple queries ("project management for 5 people") to complex enterprise architecture.

Takeaway: Domain expertise + massive data + AI = fundable business.

Source: https://startupsmagazine.co.uk/article-dragonfly-raises-ps26m-it-launches-ai-tool-simplify-software-selection

Ex-Google DeepMind founders raised $2B at $8B valuation in 12 months.

Reflection AI secured one of the largest open-source AI funding rounds ever. Brooklyn-based startup building superintelligent models to rival Meta's Llama.

• $2B Series B led by NVIDIA, B Capital, Citi
• Eric Schmidt's family office participated
• Total funding: $2.13B in just over 1 year

How they did it:
• Left with proven AI research credentials
• Positioned as open-source alternative to closed labs
• Built credible team from top AI institutions
• Attracted strategic investors needing open ecosystem
• Perfect timing: market wants AI democratization

Takeaway: Technical credibility + strategic positioning + timing = massive early funding.

https://techstartups.com/2025/10/09/top-startup-and-tech-funding-news-october-9-2025/

#startups #AI

Parallel raised $2.4M to solve startup finance chaos with AI + fractional CFOs.

Early-stage founders hate finance. Spreadsheets break, assumptions go untested, fundraising decks leave investors unconvinced.

• $2.4M seed led by Night Capital & K5 Tokyo Black
• AI agents handle forecasting and variance analysis
• Fractional CFO network provides strategic oversight
• Live integrations with QuickBooks, Xero, Puzzle

How it happened:
• Founded 2023 by ex-finance operators
• Built AI-native platform vs spreadsheet approach
• Paired automation with human CFO judgment
• Customer finished due diligence without follow-up questions

Takeaway: Finance doesn't have to be a founder nightmare when AI handles grunt work.

Source: TechBuzz News

AI presentation startup raises $30M to buy services firms — starting with founder's other company

Prezent closed a $30M Series B at $400M valuation. Twist: their first acquisition is Prezentium, co-founded by Prezent's own CEO.

• $30M raised, $74M total
• Led by Multiplier Capital, Greycroft
• Targets enterprise presentation automation

Founder playbook:
• Built AI presentation platform
• Co-founded services company simultaneously
• Used services firm as go-to-market partner
• Raised acquisition-focused round
• Merged both under one roof

Trend: AI startups buying services companies for customer bases and domain expertise.

Takeaway: Pure AI isn't enough — enterprises trust human expertise.

https://techcrunch.com/2025/10/10/prezent-raises-30-million-to-acquire-ai-services-firms-starting-with-founders-other-company/

#startups #funding

Seattle's AI2 Incubator just closed $80M Fund III to back 70 AI startups over 4 years.

Their track record: 90% of graduates raise follow-on funding, 24% get acquired (Xnor.ai to Apple, Lexion to Docusign for $165M).

Key lessons from 10 years of AI startup building:
• Avoid pure developer tools and execution-only teams without deep AI expertise
• Winners pair technical AI depth with vertical domain knowledge
• Focus on real-world applications, not "API with a coat of paint"
• Rolling admissions beat fixed cohorts for experienced entrepreneurs

They invest $600K + $1M cloud credits for 7% equity at $10M cap.

Takeaway: Domain expertise + AI technical depth = fundable startup.

Source: GeekWire

#startups #AI

Nooks just closed $43M Series B to automate sales busywork.

AI Sales Assistant Platform that automates dialing, research, and email writing for sales teams. Founded by three Stanford roommates in 2020.

Key metrics:
• $43M Series B, Kleiner Perkins led
• $285M valuation, 4x YoY revenue growth
• 70% of customer pipeline attributed to Nooks
• 2-3x call volume increase per rep

How it happened:
• Started as virtual office during pandemic (2020)
• Noticed sales teams were power users (2021)
• Pivoted to AI sales tools (2022)
• Built on GPT/Gemini, integrated with Salesforce/HubSpot
• Achieved enterprise traction, 4x revenue growth

https://www.prnewswire.com/news-releases/nooks-announces-43m-series-b-and-launches-ai-sales-assistant-platform-302285425.html

Takeaway: Sometimes the best pivot comes from watching who actually uses your product.

Brewed by Startup Beaker.

#startups #AI

Startup Beaker | Fall Incubation Has Begun 🚀

We’ve officially kicked off the Fall Batch of Startup Beaker’s incubation program — where ambitious founders turn ideas into traction.

If you’ve ever felt stuck while building your startup —
• You have the vision but not the technical muscle
• You struggle to find real users who care
• You freeze when talking to investors

Then this program is built for you.

Here’s what founders get inside:
✓ Proven frameworks to validate, scale, and raise — fast
✓1-on-1 mentoring from founders who’ve built, raised, and exited
✓ Live pitch opportunities with international VCs
✓ A global founder network that actually helps
✓ Hands-on sessions focused on execution, not theory

⚡️ Only 20 founder spots worldwide.
Next cohort starts soon.

Apply to cohort now →
startupbeaker.uz
startupbeaker.uz
startupbeaker.uz
_________

Or you can refer your friend to apply

Grant application

Join to community.
Get real traction.
Pitch global investors.

Poolside AI just closed a $500M Series B at $3B valuation — here's how two ex-GitHub executives built a coding assistant that rivals Copilot.

What it is: AI coding platform with domain-specific models trained on 500K+ open-source repos using their proprietary RLCEF method.

How it happened:
• Warner (ex-GitHub CTO) met Kant in 2017 during failed acquisition talks
• Spent 6 years plotting AI-driven developer tools while Warner was at Redpoint VC
• Founded in May 2023 with $126M seed at $526M valuation
• Hired ex-GitHub CRO Paul St. John for enterprise go-to-market
• Secured AWS partnership for Bedrock deployment
• Raised $500M Series B led by Bain Capital in 18 months

Key insight: Domain expertise + enterprise relationships + technical differentiation = massive valuation jumps.

Source: Contrary Research

Brewed by Startup Beaker.

OpenEvidence just raised $200M Series C at $6B valuation — 3 months after their $210M Series B.

What it is: AI-powered medical search engine exclusively for doctors that summarizes evidence-based research in 5-10 seconds.

• $500M total raised since 2022
• 16M clinical consultations/month (up 60% since July)
• 830% usage growth among US doctors in past year
• 15M consultations in September alone (only 1M doctors in US)
• 100M+ Americans treated by doctors using OpenEvidence this year

How it happened:
• Started 2022 with ChatGPT for medicine vision
• Secured content partnerships with AMA, NEJM, JAMA
• Built organic growth through doctor word-of-mouth
• Launched DeepConsult AI agent for advanced research
• Achieved fastest healthcare tech adoption by doctors ever

Google Ventures led the round. No tech in healthcare history has been adopted by doctors this quickly except Google itself.

Takeaway: When you solve a real pain point for professionals, organic growth becomes your best sales team.

Source: Fierce Healthcare

Brewed by Startup Beaker.

This is an incredible opportunity to develop your startup in just 4 weeks.

⚠️Applications close in just 4 days

Whether you’re at the idea stage or already building — this incubation & acceleration program is designed to take you from concept to traction with hands-on mentorship, investor feedback, and real startup execution.

Here’s what’s waiting for our founder if they join now:

📈 Practical Growth over raw concepts: You’ll work on your real startup, not theory.
💡 Expert Mentorship: Learn directly from founders who’ve built and exited companies.
🤝 Network & Funding opportunity: Connect with investors, partners, and founders across Central Asia and beyond.

Don’t miss the chance to transform your idea into a real business
👉 Apply for Incubation Grant

Serial entrepreneurs strike again with domain expertise.

Bronto just raised $14M seed to reinvent log data management for the AI era. Dublin-based startup founded by proven operators who know this space inside out.

• $14M seed led by Cercano Management
• AI-native platform handles petabyte-scale log data
• Eliminates costly tradeoffs between volume and retention
• 150+ combined years of team experience

How it happened:
• Noel Ruane: Built Voysis → sold to Apple 2020 → identified AI logging bottleneck
• Trevor Parsons: Built LogEntries → sold to Rapid7 2015 → led petabyte-scale ops
• Combined expertise spotted AI era creating 10x data volume problem
• Assembled world-class engineering team from previous companies
• Built proprietary AI-native architecture from ground up

https://www.businesswire.com/news/home/20251021830996/en/Bronto-Raises-%2414M-Seed-Round-and-Introduces-the-Worlds-Most-Powerful-Log-Data-Platform---Built-for-the-AI-Era

Takeaway: Domain expertise + proven execution track record = investor confidence in crowded markets.

Source: Business Wire

Kremniy vodiysidagi o‘zbek yuristi

Hech o‘ylab ko‘rganmisiz, nega O‘zbekistonda hali unicorn yo‘q? Yoki startapga yurist kerakmi, degan savol sizni ham qiziqtiradimi?

Bu savollarga javobni Asror Arabjonov — Clarity Flo asoschisi, Stanford Law School bitiruvchisi va hozirda Palo-Alto (AQSh)da faoliyat yuritayotgan yurist — bilan birga izladik. U Kremniy vodiysidagi hayot, erkin raqobat muhiti va o‘zbek startaperlarining global bozorga chiqish imkoniyatlari haqida o‘z tajribasini bo‘lishdi.

Podkast boshlovchilari: Laylo Xondamirovna va Shakhzod Ismoilov.

To‘liq epizodni YouTube kanalimizda tomosha qiling:
👉 https://youtu.be/SpjxDL09srQ?si=uwHEugDJ_R7aQMLy

Boardy just closed $3M pre-seed for AI voice networking.

What it is: AI assistant takes phone calls, learns what you're building, then facilitates double opt-in intros from a curated network.

The numbers:
• $3M pre-seed led by HF0
• Few thousand users in network
• Started March 2024, launched October
• 5 co-founders including ex-Clearco CEO

How it happened:
• Identified loneliness epidemic as core problem
• Built voice-first vs text-based networking
• Leveraged founder's existing investor network as seed users
• Focused on quality over quantity in network curation
• Positioned as human connection facilitator, not replacement
• Easy fundraise from existing relationships

Takeaway: Voice interfaces can differentiate in crowded markets. Start with your own network as proof of concept.

Source: TechCrunch

Brewed by Startup Beaker.