- AI Weekly
- Posts
- Google's Getting Scary Smart: Spreadsheets That Think & Chips That Crush Supercomputers
Google's Getting Scary Smart: Spreadsheets That Think & Chips That Crush Supercomputers
Google just levelled up its AI game with sheets that analyze your data and chips that make supercomputers look like calculators. ⏱️ 3 minute read
Find out why 1M+ professionals read Superhuman AI daily.
In 2 years you will be working for AI
Or an AI will be working for you
Here's how you can future-proof yourself:
Join the Superhuman AI newsletter – read by 1M+ people at top companies
Master AI tools, tutorials, and news in just 3 minutes a day
Become 10X more productive using AI
Join 1,000,000+ pros at companies like Google, Meta, and Amazon that are using AI to get ahead.
Google Puts AI in Your Spreadsheets (No Coding Required)
Remember when you needed an advanced degree to make sense of massive datasets? Those days are officially over.
Google just dropped its Gemini AI directly into Google Sheets with a simple "=AI" formula that turns anyone into a data analyst. This isn't just another spreadsheet function—it's essentially giving everyone their own personal data scientist.
Here's what it can do with just a single formula:
• Sort through emotions: Type =AI("Classify this sentence, as either positive or negative.", A2)
and watch as your customer feedback gets instantly categorized • Summarize paragraphs: Need the TL;DR version of that novel-length cell? Just use =AI("Summarize this info in 1 very short sentence", A2)
• Categorize anything: Sports teams, product types, complaint categories—just ask and Gemini delivers
The implications are huge. Tasks that once required specialized Excel wizards can now be handled by anyone with basic spreadsheet knowledge. That promotion-worthy analysis you spent days on? Your intern might now complete it before lunch.
To get started, you'll need a Google Workspace account or Google One AI Premium subscription. Look for the "Ask Gemini" button in your sheets or just start typing "=AI" in any cell.
Ironwood: Google's New AI Chip Makes Supercomputers Look Like Toys
While Microsoft and OpenAI battle NVIDIA's GPU shortages, Google just casually built a chip that makes the world's fastest supercomputer look like your grandma's flip phone.
Google's new Ironwood chip—the seventh generation of their custom Tensor Processing Units (TPUs)—is built specifically for AI inference (the part where AI actually gives you answers). And the numbers are mind-boggling.
When combined into a full system:
• It delivers 42.5 exaflops of computing power—that's 24 TIMES more than El Capitan, currently the world's fastest supercomputer • Each chip packs 192GB of high-bandwidth memory—6X more than Google's previous generation • It's nearly 30 TIMES more power-efficient than Google's first TPU from 2018
Why does this matter? While companies like Microsoft and Meta wait in line for NVIDIA's chips, Google controls its own destiny. They've built the perfect hardware for running their Gemini AI models at massive scale without astronomical energy bills.
This isn't just about raw power—it's about making AI inference (which happens billions of times daily) faster and cheaper than ever before.
The Big Picture: Google's Two-Pronged AI Strategy
Google isn't just building impressive technology—they're executing a clear strategy:
Democratize AI with tools like Gemini in Sheets, putting advanced capabilities in everyone's hands
Own the infrastructure with custom chips like Ironwood that give them independence from suppliers and lower operating costs
Together, these moves position Google to deploy increasingly sophisticated AI models while actually reducing costs—something their competitors are struggling to figure out.
Executive Summary
• Google has integrated Gemini AI into Google Sheets with a simple "=AI" formula that can categorize, analyze sentiment, and summarize data • Their new Ironwood AI chip delivers 42.5 exaflops—24 times more powerful than today's fastest supercomputer • Each Ironwood chip packs 192GB of memory (6x their previous generation) and uses half the energy • While competitors rely on NVIDIA's chips, Google's in-house approach gives them control over both cost and supply
What's next? With Google simultaneously making AI more accessible to users and more efficient to operate, how long until even the most specialized knowledge work gets transformed by these tools? Your spreadsheet skills might still be valuable—but for how much longer?


Reply