Hi {{first_name}}!
This week, Larry Ellison briefly became richer than Elon Musk by signing the most expensive IOU in tech history, Microsoft started cheating on OpenAI with Anthropic in the most expensive love triangle ever, and everyone discovered that stealing books for AI training costs exactly $3,000 per pilfered paperback.
As we continue to build at Context Ventures, we've launched Startup Intros with a simple mission: to help early-stage founders find the right investors, faster and smarter.
🔥 Context VC & Startup Intros Events Coming Up
Check it out, stay plugged in, and don’t miss these hot upcoming events:
Sep 24: Founder Real Talk: Beyond VCs: We dive into alternative funding, including bank lending, SBA loans, grants, and bonds. Sign up on Luma!
Sep 29: Fintech Pitch Night: Applications to pitch are open! Join us in SF for an exciting pitch night for Fintech. Sign to pitch here!
Oct 6: Startup Intros Product Launch Party: Celebrate our platform launch with founder tools demo and SF's best startup networking. Apply on Luma!
Oct 9: Early-Stage Matchmaking: Curated founder-investor matching during SF Tech Week. High-signal connections for serious builders. Apply on Luma!
Oct 22: Live Podcast w/ Meka Asonye: First Round Capital's Partner talks about GTM, founder support, and building from Day 0. Sign up on Luma!
Keep reading for the latest edition of Trending Thursday!
🎯 BLUF: Bottom Line Upfront
Oracle's $300B OpenAI Moonshot: Larry Ellison signs history's largest cloud deal, stock jumps 36%, briefly becomes world's richest man while canceling employee raises to pay for it
Microsoft's Anthropic Affair: Ditches OpenAI for Claude in Office 365, paying Amazon to access better AI while negotiating messy breakup terms
The $1.5B Copyright Bill Comes Due: Anthropic settles for $3,000 per stolen book, setting a precedent that could bankrupt half of AI
Everyone's Building Chips Now: OpenAI drops $10B on Broadcom partnership, Google forcing TPUs everywhere, Nvidia secretly renting its own GPUs
Apple's Desperate AI Scramble: Building "World Knowledge Answers" search engine while testing every AI model that will return their calls
🚀 Oracle's $300B Bet: When Boring Enterprise Software Becomes AI's Sugar Daddy
💰 The Deal That Broke Mathematics (And Made Larry Richer Than Elon)
Oracle just signed a $300 billion compute deal with OpenAI that's so large it defies comprehension. This isn't just the biggest cloud deal ever: it's bigger than the GDP of Finland.
The Absolutely Bonkers Numbers:
$300B over 5 years: Starting 2027, for AI compute (that's $164M per day)
Stock explosion: Up 36% in one day, biggest jump since dot-com era
Larry's net worth: Briefly passed Elon at $383B (41% stake in Oracle)
2030 projection: $144B in cloud revenue (from $3.3B today—a 43x increase)
Market cap: Hit $922B, approaching the trillion-dollar club
The Employee Sacrifice: To fund this AI transformation, Oracle is eliminating cash raises and bonuses, offering stock grants instead. Because nothing says "we believe in our future" like making employees bet their grocery money on it.
The Infrastructure Insanity:
Building 4.5 gigawatts of data centers (enough to power Microsoft)
Capex jumping from $21B to $35B this year alone
Signed $317B in total contracts last quarter (didn't name customers, but we can guess)
Why This Matters: Oracle just went from the company your dad's IT department uses to the kingmaker of AI. They're essentially becoming OpenAI's landlord, and just speedran the entire cloud revolution in one earnings call.
💔 The Microsoft-OpenAI Divorce is Getting Messy
Microsoft is bringing Anthropic's Claude into Office 365, marking the beginning of the end of its OpenAI exclusivity. The twist? They're paying Amazon to access the models.
The Breakup Details:
Claude for Office: Better at Excel automation and "more aesthetically pleasing" PowerPoints
The AWS Humiliation: Microsoft paying rival Amazon for Anthropic access
OpenAI keeps basics: Still powers general Copilot features (for now)
GitHub already switched: Previously moved to Anthropic for advanced features
The Strategic Hedging:
Testing Google's Gemini for potential deals
Building own MAI-1 models internally
Already using xAI models in some products
Basically dating everyone while still technically with OpenAI
The Timing: This comes as Microsoft and OpenAI negotiate over the startup's IPO plans and equity arrangements. Nothing says "healthy relationship" like secretly shopping for your replacement during contract negotiations.
Bottom Line: Microsoft's Copilot is heading for $1B+ revenue, but adoption has been throttled by quality issues. Users complained about paying $30/month for buggy spreadsheet help. The Anthropic switch suggests Microsoft values results over relationships, and Claude delivers where GPT doesn't.
📚 The $1.5B Antropic Copyright Reality Check
💸 Anthropic Learns That "Move Fast and Steal Things" Has a Price Tag
Anthropic just agreed to pay $1.5 billion to settle copyright claims, setting a precedent that should terrify every AI company that trained on Library Genesis.
The Expensive Education:
$3,000 per book: For 500,000 pirated books in the class action
Largest copyright settlement: In U.S. history (previous record wasn't even close)
The smoking gun: Downloaded from Library Genesis and Pirate Library Mirror
Judge's ruling: Training on purchased books = fair use, pirated books = theft
The Industry Bombshell: Court filings revealed employees at OpenAI and Meta used the same pirated datasets. With 40+ similar lawsuits pending, Anthropic's settlement becomes the template. Everyone's about to get very expensive bills.
📜 The New Licensing Reality from Reddit and Co.
Publishers are launching Real Simple Licensing (RSL), backed by Reddit, Medium, and Yahoo:
Pay-per-crawl or pay-per-inference models
Automated licensing negotiations
Industry solidarity against free scraping
The end of the "it's on the internet, so it's free" era
Why It Matters: AI companies have been playing a game of chicken with copyright law, betting they'd be too big to sue by the time anyone noticed. Anthropic just blinked first, and at $3,000 per stolen book, the entire industry's training data strategy just became a liability worth hundreds of billions.
🎯 The Custom Chip Gold Rush: Everyone Wants Their Own Silicon
The New Chip Independence Movements: The entire industry just decided that relying on Nvidia is for suckers:
💎 OpenAI's $10B Broadcom Deal: Mass-producing custom AI chips starting 2026, joining Google/Amazon in the "build your own" club. Internal use only (for now). Still desperately needs Nvidia chips while building alternatives, the definition of hedging your bets.
🌐 Google's TPU Colonial Expansion: Approaching Nvidia-focused clouds about hosting TPUs, already struck deal with Fluidstack for New York data center. Courting CoreWeave and Crusoe (both Nvidia strongholds). Message clear: every TPU in a rack is one less Nvidia GPU.
🔄 Nvidia's $1.5B Self-Rental Scheme: Quietly paying Lambda to rent its own chips back. Lambda is already an Nvidia customer and investment. The most circular financial arrangement in tech, basically paying itself to juice demand numbers.
🎯 Nvidia's First Inference Chip: Launching Rubin CPX specifically for inference workloads, available end of 2026 (same timeline as everyone else). Cursor, Runway, Fireworks already signed up. Finally, admitting training and inference need different silicon.
Why It Matters: The chip wars have entered their final form: vertical integration or death. Every major AI player is designing custom silicon because Nvidia's 80% margins are too juicy to let stand. Eventually, the question won't be "whose model are you using?" but "whose chips are you stuck with?"
⚡ Quick Hits: When Everything Else is Also Insane
🍎 Apple's AI Panic Mode: Building "World Knowledge Answers" search engine for Siri, testing Google's Gemini and Anthropic's Claude because their own models are trash. That $20B/year Google search deal looking shaky. Also announced iPhone Air at 5.6mm, so thin it needs external battery for "all day" use.
🚀 SpaceX's $17B Spectrum Grab: Buying EchoStar's wireless spectrum to enable direct satellite-to-phone service. Starlink can now beam internet to regular smartphones without special hardware. T-Mobile's exclusivity dead, Boost Mobile gets access. Global cell coverage from space incoming.
🤖 Tesla's Robot Hand Crisis: Musk admits Tesla "struggling" with Optimus robot hands, which represent "majority of engineering difficulty." Still claiming $20-25K manufacturing cost eventually. Goal of 5,000 robots in 2025 looking impossible. Meanwhile, proposing $1 trillion pay package requiring $8.5T market cap.
💰 X Money Going Nowhere: Elon's payments ambitions stalled by New York regulators. CEO quit, two compliance chiefs fled, multiple executives gone. The "everything app" dream dying while Musk focuses on xAI instead.
📝 Nvidia vs Congress on China: Nvidia and White House AI czar David Sacks calling lawmakers "doomers" for wanting to restrict China chip sales. Congressional hawks firing back. The geopolitical chip war getting personal.
💸 What Other Startups are Popping Off
Here's a roundup of this week's trending startup activity:
🤖 AI Agents & Automation
CREAO: Infinitely personalized software for every team member (finally, software that gets you)
YouMind: The first AI creation studio where learning meets writing (your thoughts just got an upgrade)
Latitude 2.0: The 1st agent that builds agents (inception, but for AI agents)
Incredible Small 1.0: Agentic AI Models that can take 1000+ actions at once (multitasking on steroids)
Bika.ai: Chat, build, manage one-person company (the ultimate solopreneur sidekick)
Spiral: Analyze your reviews & support data with AI (customer feedback finally makes sense)
🛠️ Developer Tools & Infrastructure
Incerto: The AI Copilot for Databases (databases that actually speak human)
Shotgun CLI: Codebase-aware spec engine for Cursor, Claude Code & Lovable (specs that know your code better than you do)
Web to MCP: Send any website components to Cursor or Claude via MCP (copy-paste, but make it intelligent)
EasyCode: Build serious apps with AI. 10x credits. No code required (coding without the crying)
100 Vibe Coding: From zero to your first project in 100 challenges (learn by doing, but with style)
Solid: AI that builds real web apps (no more "coming soon" pages)
Vectorize 2.0: Complete RAG agents (chatbot, MCP) with little or no code (RAG without the drag)
Empromptu: Describe your idea to build AI apps that actually work (ideas that turn into reality)
Ona (formerly Gitpod): Sandboxed parallel coding agent platform (coding agents in their own playground)
Gram by Speakeasy: Instantly create MCP servers and SDKs that LLMs understand (APIs for the AI age)
Billing SDK: Open source billing UI components (because billing doesn't have to be boring)
🎨 Creative & Content Tools
maestro by beatoven.ai: Our latest music generation model, trained on licensed data (AI music that won't get you sued)
VoxDeck: 3D charts, motion cover & avatar generation for slides (presentations that actually present)
PhotoFox AI: Turn any product photo into 100+ on-brand assets in minutes (one photo, endless possibilities)
CapCut AI Suite: Create, edit, or remix content with AI in a simple editor (video editing for the rest of us)
FunBlocks AI Slides: Instant AI Slides, Smooth Editing with Markdown and AI Help (PowerPoint just got out-powered)
Higgsfield Ads 2.0: Product placement is finally solved by Higgsfield (subtle ads that aren't so subtle)
Tripo AI 3.0: Create AI 3D model with text and image in seconds for free (3D modeling at the speed of thought)
VidAU - AI Video: Turn Product Images into Scroll-Stopping Video Ads (static photos just got animated)
🏢 Sales & Marketing Tools
ClearCRM: Comprehensive CRM Suite with built-in automation builder (CRM that actually helps you close deals)
Kaylin AI: Full-service AI powered marketing for small businesses (marketing without the marketing degree)
SEO Mega Report: GPT-5 all-in-one SEO report for Google & ChatGPT (SEO reports that don't put you to sleep)
Genstore.ai: Start selling online with just a prompt (e-commerce in a sentence)
💼 Business & Productivity
Meeting.ai: an AI notetaker built for visual learners (meeting notes with pictures, finally)
Speech to Note Mobile App: Taking Notes on the Go! Online, offline (notes that follow you everywhere)
Katalog: Talk to your reading list (your books can finally talk back)
Uxia: Validate your User flows UX & UI in seconds with AI (UX validation at light speed)
Dockify: Customize and switch your Mac dock instantly (because your dock should match your mood)
Spiky Mobile App: Your meeting assistant, now in your pocket (meetings that don't suck, on mobile)
How 433 Investors Unlocked 400X Return Potential
Institutional investors back startups to unlock outsized returns. Regular investors have to wait. But not anymore. Thanks to regulatory updates, some companies are doing things differently.
Take Revolut. In 2016, 433 regular people invested an average of $2,730. Today? They got a 400X buyout offer from the company, as Revolut’s valuation increased 89,900% in the same timeframe.
Founded by a former Zillow exec, Pacaso’s co-ownership tech reshapes the $1.3T vacation home market. They’ve earned $110M+ in gross profit to date, including 41% YoY growth in 2024 alone. They even reserved the Nasdaq ticker PCSO.
The same institutional investors behind Uber, Venmo, and eBay backed Pacaso. And you can join them. But not for long. Pacaso’s investment opportunity ends September 18.
Paid advertisement for Pacaso’s Regulation A offering. Read the offering circular at invest.pacaso.com. Reserving a ticker symbol is not a guarantee that the company will go public. Listing on the NASDAQ is subject to approvals.
🌟 Editor's Note
At Startup Intros, our mission is to bring the latest founder-investor news straight to your inbox, keeping you ahead in the fast-paced world of Silicon Valley. Our free weekly newsletter delivers curated insights, deals, and trends to help you navigate the startup ecosystem.
Stay tuned for our upcoming paid daily newsletter and fundraising app to help you close your next round!
💭 Parting Thoughts
This week proved that Silicon Valley's new business model is "become your own supplier." Oracle's betting $300 billion it can be the AWS of AI, Microsoft's cheating on OpenAI with every model that texts back, and literally everyone is designing chips because Nvidia's margins are too good to let stand.
The winners won't be the ones with the best models. They'll be the ones who own the chips, the data centers, the power plants, and apparently, the pirated book settlements.
Forward to a friend or hit reply to let me know what you're seeing in your world.
Forward to a friend and hit reply to let me know what you're seeing in your world.
Till next time!

Dev Chandra
CEO @ Startup Intros
Associate @ Context VC
LinkedIn: /in/devchandra

Tim Hsia
Investor @ Context VC
Co-Founder @ Startup Intros
LinkedIn: /in/timhsia
P.S. Raised this week, and we missed you? Want to be featured? Have tips or funding questions? Reply or DM us as we’re here to help.
Startup Intros Trending Thursday: Your trusted source for founder-investor insights, delivered with clarity and focus.