Good morning from San Francisco,

After 17 years of research, both Microsoft and Crunchbase have emerged with something remarkable. One wants to stabilize quantum particles, the other wants to predict startup destinies. Both remind us that patience in tech isn't just a virtue – it's a strategy.

Microsoft has finally managed to wrangle some Majorana particles into submission, creating a quantum chip that could scale to a million qubits. It's like they've been trying to teach physics a new party trick, and physics finally agreed to cooperate.

Meanwhile, Crunchbase has transformed from a startup encyclopedia into a startup fortune teller. Their AI can now predict major deals with surprising accuracy, though it's still working on its "startup obituary" skills.

Enjoy reading,

Marcus Schuler

The Long Game: How Microsoft's Quantum Gamble Could Pay Off Big

After a 17-year research marathon, Microsoft has unveiled its Majorana 1 processor. It's a quantum chip that could pack a million qubits into a space barely larger than today's desktop CPU.

The secret sauce? A new "topoconductor" material. It harnesses the Majorana particle, first theorized in 1937. Microsoft’s approach is more stable than traditional quantum computers with their fragile qubits. Think of it as balancing a pencil on its tip versus laying it flat.

Microsoft took the road less traveled. While IBM and Google worked with standard qubits – about as stable as a house of cards in a hurricane – Microsoft created an entirely new material. It's a unique blend of indium arsenide and aluminum designed to wrangle Majorana particles into doing useful quantum tricks.

"After 17 years, we are showcasing results that are not just incredible, they're real," says Zulfi Alam, Microsoft's VP of quantum. This is fortunate because explaining imaginary results to shareholders would have been awkward.

The current prototype has eight topological qubits. But the architecture could scale to a million. That's like going from a calculator to a supercomputer in one leap.

Microsoft's quantum team approached this like true pioneers. "We took a step back and said, 'Ok, let's invent the transistor for the quantum age,'" says technical fellow Chetan Nayak. Solving today's problems wasn't challenging enough.

DARPA has noticed. They've picked Microsoft as one of two companies for their quantum computing program's final phase. The goal? A fault-tolerant quantum computer "in years, not decades." However, in quantum terms, that timeline might be simultaneously both.

Why this matters:

• Microsoft may have found quantum computing's holy grail – stability without sacrificing power. Like finally teaching a cat to sit still... while juggling
• By solving the materials problem first, Microsoft proves that sometimes the longest way around is actually the shortest way home
• This could be to quantum computing what the transistor was to classical computing – though hopefully with fewer "turn it off and on again" moments

Read on, my dear:

• implicator.ai: Microsoft's 17-Year Quantum Gamble Pays Off (Finally)
• Microsoft: Microsoft’s Majorana 1 chip carves new path for quantum computing
• The Verge: Microsoft announces quantum computing breakthrough with new Majorana 1 chip

AI Photo of the Day

Prompt:
Photography, girl with black hair, pink outfit, pink flamingos, high fashion, I is colors, big city

AI-Powered Oracle: How Crunchbase Became Silicon Valley's Fortune Teller

After 17 years of cataloging startup histories, Crunchbase is now predicting their futures. The market intelligence firm has transformed its vast database into an AI-powered crystal ball.

The secret sauce? It's not just public data anymore. The system analyzes digital footprints from 80 million users. When startup employees update their profiles or investor interest spikes, Crunchbase's AI smells money in the air.

And it's surprisingly accurate. The system spotted Anthropic's $2 billion fundraise before the ink dried. It even called Coda's Grammarly acquisition with 93% accuracy. Predicting startup failures remains its Achilles' heel - those predictions are right less than half the time. It turns out zombie startups can shuffle along for surprisingly long.

Getting here wasn't cheap. CEO Jager McConnell made the tough call to cut a third of his staff - primarily sales and marketing - to build the technical muscle. The old business model was already wheezing under pressure from chatbots that could dig up similar data.

"Historical data companies are already dead," McConnell says cheerfully, as if someone has already seen the future. Speaking of which - when Crunchbase turned its predictive powers on itself, it saw an acquisition coming. McConnell doesn't disagree.

Why this matters:

• In the ChatGPT era, knowing yesterday's news is old news - the real value is in predicting tomorrow's headlines
• Your users' behavior patterns might be worth more than the data they're looking at
• The best time to reinvent yourself is before you have to - just ask the fortune tellers at Crunchbase

Read on, my dear:

• implicator.ai: Database Giant Crunchbase Transforms into Tech Industry Prophet

🇪🇺 Europe's AI Regulation Timeline: The State of Play - From Madrid to Malta: Who's Ready to Police AI?

The European Union's AI Act enforcement begins in August 2025, creating a scramble among member states to establish their regulatory frameworks.

Here's where the 27 EU nations currently stand:

🏃‍♂️ Early Movers:

• 🇪🇸 Spain leads with AESIA, established in September 2023, already conducting preliminary investigations
• 🇲🇹 Malta has designated MDIA and IDPC as joint regulators, ready to begin enforcement by mid-2025
• 🇩🇰 Denmark has appointed its Digital Government Agency as coordinator, with full enforcement planned by early 2026

⚙️ Building Momentum:

• 🇩🇪 Germany expects to finalize its draft law by early 2025, with Bundesnetzagentur as the lead agency
• 🇫🇷 France aims for mid-2025 designation of multiple bodies, including DGE and CNIL
• 🇵🇱 Poland plans to have its new AI Commission operational by August 2025
• 🇮🇹 Italy's ACN awaits legislative approval by mid-2025

🚦 Still at the Starting Line:

• 🇦🇹 Austria, 🇧🇪 Belgium, and 🇧🇬 Bulgaria must designate authorities by August 2025
• 🇳🇱 Netherlands is coordinating 22 advising agencies, targeting mid-2025 for framework completion
• 🇸🇪 Sweden debates between IMY and DIGG, with designation expected by mid-2025
• 🇷🇴 Romania and 🇸🇰 Slovakia are creating new authorities from scratch, aiming for late 2025 readiness

⚠️ Why this matters:

• With only 7 months until the August 2025 deadline, most EU countries are still in planning stages
• The gap between early adopters and latecomers could create an uneven enforcement landscape in early 2026
• Countries starting late may face rushed implementations, potentially compromising regulatory effectiveness
  
  Source:
  - Kevin Schawinski

AI & Tech News

Silicon Valley's Power Players Hit Popularity Low

A new Pew Research survey reveals a stark decline in public opinion of tech leaders Elon Musk and Mark Zuckerberg, with both facing majority unfavorable views from Americans. While Musk maintains strong Republican support, particularly after joining the Trump administration, Zuckerberg faces bipartisan disapproval, with two-thirds of Americans viewing him unfavorably.

Zuckerberg's AI Lobbying Mission

Mark Zuckerberg visited the US Capitol to discuss artificial intelligence regulation with senators, coming on the heels of Meta's announced $65 billion investment in AI technologies and amid growing tensions over European tech regulations. The Meta CEO's timing is significant, coinciding with his company's recent policy shifts and increasing alignment with the current administration's stance on tech regulation and social policies.

Social Media Giants Show Far-Right Bias in German Elections

TikTok and X's algorithms heavily favor Germany's far-right AfD party in recommendation feeds, with TikTok showing 78% pro-AfD content despite the party polling at just 20%. According to new Global Witness research, X follows with 64% right-leaning recommendations. These findings emerge from test accounts that followed all major German political parties equally, suggesting the platforms' algorithms may be inadvertently shaping political discourse ahead of Sunday's federal election.

AI Startup Helps Old Machines Run Smarter

Augury raised $75 million to expand its AI technology that monitors factory equipment for repairs. The system uses sensors to track machine vibrations, sound, and temperature, building what CEO Saar Yoskovitz calls "the malfunction dictionary" from over half-a-billion hours of operational data. Major manufacturers like PepsiCo and Nestle are already on board.

Google's Digital Einstein: The Future of Scientific Discovery

Google's revolutionary AI lab assistant has achieved in mere days what typically takes human scientists years of painstaking research. Working alongside top institutions like Stanford and Imperial College London, this digital researcher correctly identified patterns in antimicrobial resistance that matched years of human-led discoveries.

Google Pulls Gemini from iOS App, Pushes Users to Standalone App

Google is dropping Gemini from its main iOS app. Users will need to download the dedicated Gemini app instead. Simple as that. The standalone Gemini app offers more features, including Gemini Live, which was never available in the main Google app.

Chinese AI Model Breaks Free From Censorship

Perplexity AI has unveiled R1 1776, an uncensored version of its DeepSeek language model that tackles topics previously restricted by Chinese censorship. The open-source release maintains the model's strong reasoning capabilities while freely addressing sensitive topics, marking a small revolution in AI transparency.

AI Decoded 🔓

Your guide to mastering AI tools - no tech degree required.

From "Help!" to "Got It": ChatGPT's Excel Mastery Guide ✨

Remember the days when Excel felt like a maze of formulas and hidden features? ChatGPT has changed the game. Think of it as having a patient Excel expert who never gets tired of explaining VLOOKUP for the hundredth time. 🧙‍♂️

Here's your roadmap to Excel mastery with ChatGPT:

• Set the Scene Like a Pro 🎬
• Tell ChatGPT your role: "I'm a marketing manager working with customer survey data"
• Describe your Excel version and skill level
• Explain what you're trying to achieve, not just what's broken
• Ask for Formula Help (The Smart Way) 🔍
• Request step-by-step breakdowns of complex formulas
• Ask for explanations in plain English
• Get alternatives when a formula isn't working: "Is there a better way than VLOOKUP?"
• Transform Data Cleaning from Pain to Gain ✨
• Get Power Query scripts for automatic data cleanup
• Learn how to split, merge, and restructure data efficiently
• Handle those pesky irregular formats and missing values
• Automate Like a Boss 🤖
• Request VBA scripts that turn hours of work into seconds
• Get code for automatic dashboard updates
• Learn how to make your spreadsheets refresh themselves
• Level Up Your Visualizations 📊
• Create dynamic dashboards that update automatically
• Build interactive charts with slicers
• Design report layouts that actually make sense
• Become an Error-Hunting Expert 🕵️‍♂️
• Turn #VALUE! and #REF! errors into learning opportunities
• Get multiple solutions for common spreadsheet problems
• Learn to prevent errors before they happen

Try these proven prompts: 💡

"I'm a marketing analyst working with Excel 365. I have customer names in column A (format: 'LastName, FirstName'). Write a formula that splits these into separate columns, accounting for middle names and titles like 'Dr.' or 'Mrs.'" 📝

"Write a VBA script that loops through all worksheets, finds columns with 'Revenue' in the header (row 1), and creates a summary sheet with monthly totals. Include error handling and progress updates." 🔄

"I have a VLOOKUP formula that's returning #N/A errors: =VLOOKUP(A2,Sheet2!B:D,2,FALSE). Walk me through potential causes and fixes, then suggest a more robust alternative using INDEX-MATCH." 🔧