Google Just Hit the Turbo Button on AI

Welcome back apprentices! 👋

Thought Google was just napping while OpenAI, Microsoft, and every ChatGPT-wrapped startup stole the AI limelight? Not even close. 

At Cloud Next 2025, Sundar Pichai didn’t just drop a few shiny features — he basically rolled out a masterplan to turn Google Cloud into the central nervous system of enterprise AI.

We’re talking AI in your inbox, your firewall, your dev tools, your dashboards — your everything. This wasn’t a keynote, it was a power move. A signal that while others were showing off what AI can do, Google’s been quietly building the stack it will run on.

And let’s be real: it kind of feels like they’ve been engineering the AI equivalent of the Iron Man suit — while everyone else is still stitching together chatbots with duct tape.

In today's email

• $4.8 Trillion Boom Could Deepen Global Inequality

• Shopify CEO to Staff: Use AI Before You Hire

• OpenAI Eyes $500M+ Acquisition

• Google’s AI Grand Slam

• Stanford’s 2025 AI Report

• Even more AI magic

• Test the prompt (NEW)

_{Read Time: 4 minutes}

Quick News 

🌍 The UN just dropped a spicy report predicting AI will balloon into a $4.8 trillion market by 2033 — basically the GDP of Germany, but with fewer sausages. The catch? Most of the world isn’t invited to the AI party. Just 100 companies (mostly from the U.S. and China) hold the keys to the kingdom, while 118 countries are still stuck on dial-up and getting ghosted from AI policy talks. Oh, and up to 40% of jobs might be “reimagined” into oblivion.  AI is climbing the economic ladder two rungs at a time, while half the world’s still looking for the instruction manual. 

📈 In a company-wide memo, Shopify CEO Tobi Lütke declared AI the new baseline—like email, but smarter and with zero coffee breaks. He told teams to use AI as a co-pilot, tutor, and tireless intern before asking for more humans. From prompt engineering to automated workflows, the message is clear: if you’re not using AI to do 10x the work, you might just be working like it’s 2019. And before you draft that job req, ask yourself: could a robot do it better — and quietly?

📱 OpenAI is reportedly in talks to acquire io Products, a stealth AI hardware venture co-founded by legendary Apple designer Jony Ive and backed by CEO Sam Altman. The startup is working on screenless AI devices—think “AI phone” without the phone—built by former Apple hardware leads and powered by OpenAI’s tech. With a $1B fundraising target and a potential $500M+ valuation, this project could be the company's bold leap into consumer hardware.

Google 
Hardware, Software, and World Domination (Probably)

At Cloud Next 2025, Google didn’t just show up — they made it loud, fast, and undeniably futuristic. With the energy of a company that’s tired of hearing “but OpenAI,” Sundar Pichai took the stage to outline a vision where AI is not just a feature — it’s the foundation. 

From enterprise copilots to custom chips and smarter security, Google made its play to own the full AI stack, and honestly, it looks like they brought the receipts.

Major Announcements at Google Cloud Next 2025:

🔩 Ironwood TPUs – Google’s newest AI chips with 2.3x better compute-per-watt than v5p, built for massive scale and even bigger models.

🧠 Gemini 2.5 – A serious upgrade boasting 1M+ token context windows and multimodal mastery — ideal for enterprise-grade reasoning and memory.

🧰 Vertex AI Agent Builder – Drag-and-drop your way to powerful AI agents that understand workflows like your best project manager.

💻 Gemini Code Assist – Google’s take on AI pair programming — enterprise-ready and deeply integrated across Cloud.

🔐 Gemini in Security Command Center – Let AI find threats before they find you. A brainy upgrade for your cyber defense.

🌐 Model Garden expansion – Welcoming Claude, Llama 3, Mistral and more — because Google wants to be the mall for all things AI.

⚡ Gemini 1.5 Flash – A speedy, cost-efficient Gemini flavor built for real-time tasks and ultra-responsive AI apps.

📎 Gemini in Workspace & ChromeOS – Now your docs, slides, and even your Chromebook are thinking alongside you.

🤝 Big-name Partnerships – New and expanded alliances with SAP, Hugging Face, Typeface, NVIDIA, and more to scale AI globally.

This isn’t just an AI glow-up — it’s Google going full Tony Stark. They’ve got the tools, the hardware, and the ambition to be the backbone of the AI-powered enterprise. 

So what’s the deal? Google is redefining the terms of the AI race. With custom silicon, frontier models, and enterprise-grade AI tools stitched into every layer of its cloud stack, the company is building what looks like the first truly end-to-end AI ecosystem. 

While competitors are still packaging AI as features or standalone apps, Google is rewriting the operating system of modern business. Think less chatbot, more cloud-wide intelligence — deployed at scale, embedded by design, and backed by some of the most mature tooling in the game.

Google’s message is clear: we’re not here to play, we’re here to build the infrastructure of the AI-native enterprise.

Stanford
The State of AI in 2025: Smarter, Cheaper, Still a Little Chaotic

Stanford just dropped its AI Index 2025, and it’s basically the annual checkup for the entire AI industry. Spoiler alert: the patient is thriving, but also maybe stress-eating compute and skipping safety checkups.

AI in 2024 became way smarter, dramatically cheaper to use, and nearly impossible to avoid if you're running a business. But it also got pricier to build, harder to govern, and... let’s just say the safety net is more like a suggestion box.

From Cool Side Project to That One Employee Who Does Everything

Last year, AI officially stopped being the intern and started running entire teams. 

Adoption exploded:

• 78% of companies now use AI (up from 55% in 2023)
  

• 71% are using generative AI (up from "just messing around with it" last year)
  

• Productivity is up in areas like customer support and dev workflows
  

• But the ROI glow-up is still pending: most businesses are seeing <10% cost savings and <5% new revenue
  

Translation: AI’s in your business. Whether it’s helping or just looking busy depends on how you use it.

Here’s the twist: while using AI models is now 280x cheaper than it was two years ago, building them could fund a small space program.

• GPT-3.5-level output now costs just $0.07 per million tokens (down from $20)
  

• Meta’s Llama 3.1? $170 million to train — and that’s just the warm-up
  

• Training compute doubles every 5 months
  

• Oh, and 48% of the internet is now blocking AI crawlers like it’s fighting off a data apocalypse
  

AI is affordable to run, wildly expensive to invent, and increasingly hard to feed.

Smarter Than Ever (Still Not Ready to Run the World)

2024 saw a ridiculous leap in model capabilities. On coding tasks (SWE-bench), models jumped from 4.4% → 71.7% in one year. AI even beat humans in competitive math.

But on "Humanity’s Last Exam" — basically the SAT for robots — models scored just 8.8%. So yes, AI can fix your code and do your taxes, but don’t ask it to explain your dreams or run a board meeting. Yet.

The U.S. is still ahead, but China’s catching up fast. The model performance gap shrunk from 9% to just 1.7% in a year. Meanwhile, open-source models are now competing head-to-head with the big locked-down ones.

For businesses, that means more options, shorter release cycles, and — let’s be honest — a lot more FOMO.

Safety: The Part Everyone Keeps “Planning to Get To”

Stanford’s report throws some polite shade here: AI incidents rose 56% last year, but actual safety regulation? Not so much.

• U.S. passed 4 AI laws (but proposed 221 — classic)
  

• Standard safety testing is rare
  

• Corporate accountability? Still catching up with “move fast and break things” vibes
  

Europe’s trying to play grown-up with the EU AI Act. The rest of the world is still arguing over who’s in charge of the off switch.

So what’s the Deal? Stanford’s AI Index isn’t hype — it’s a cold, data-packed reality check. It’s not trying to sell you a product; it’s showing you the map. And right now, that map says: AI is accelerating, access is broadening, governance is lagging, and the frontier is getting expensive fast.

This report is quietly screaming that we’re building civilization’s next operating system without a QA team. So yes, it’s time to use AI — but also to seriously think about who’s steering, who’s watching, and what happens if we hit a bug at scale.

And while AI powers more of your stack, it’s also becoming:

• More expensive to scale
  

• Harder to regulate
  

• And still capable of saying weird things in a customer chat at 2 a.m.

Help Your Friends Level Up! 🔥

Hey, you didn’t get all this info for nothing — share it! If you know someone who’s diving into AI, help them stay in the loop with this week’s updates.

Sharing is a win-win! Send this to a friend who’s all about tech, and you’ll win a little surprise 👀

Even Quicker News 

⏳ Daniel Kokotajlo, ex-OpenAI and now full-time AGI alarm bell, says machines could outcode, out-think, and outvote us by 2027 — seatbelts optional. And by 2029, superintelligent AI may be running the global economy — efficient, tireless, and still not asking for health benefits.

📷 Google’s Project Astra is here to let your phone see the world and sass you about it in multiple languages—kind of like Google Lens with a personality and caffeine. It’s not full AI sidekick mode yet, but smart glasses better start watching their back.

🕵️ Metro Bank just launched a scam-sniffing AI that lets you snap a pic of shady emails and get the truth via WhatsApp — like having Sherlock Holmes in your pocket, but faster and with emojis. Bonus: it even tattles to the authorities for you.

Today’s Toolbox

🤝 Google just launched A2A, the universal handshake for AI agents—so your HR bot can finally talk to your finance bot without starting a Slack war. It’s like diplomacy for robots, and it might just save your workflows (and your sanity).

💼 Microsoft just gave its Copilot a memory, a personality, and the power to book your dinner plans — basically, it’s your new overachieving AI BFF. Get ready for the assistant who remembers your dog’s birthday and finds you concert tickets.

🎨 MidJourney just dropped its first major update in over a year — and it’s like your AI got a makeover, hit the gym, and learned photography. V7 delivers 30% faster renders, ultra-realistic lighting, and keeps all your old sref styles in check.

🧪 Test the Prompt

A playground for your imagination (and low-key prompt skills).

Each send, we give you a customizable DALL·E prompt inspired by a real-world use case — something that could help you in your business or job if you wanted to use it that way. But it’s also just a fun creative experiment.

You tweak it, run it, and send us your favorite. We pick one winner to feature in the next issue.

Bonus: you’re secretly getting better at prompt design. 🤫

The winner is…

Last send, we challenged you to test GPT-4o’s visual generation skills with this prompt.

Here’s the WINNER:

Congrats to Brandon from Sacramento!🥳

Want to be featured next? Keep those generations coming!

🎨 Prompt: “Product Demo from a Parallel Timeline”

“A high-stakes product demo on a floating stage above a [unrealistic terrain], hosted by a hybrid presenter who is part [animal] and part [tech object]. The presenter wears a suit patterned with [random motif] and unveils a glowing, levitating product. The product displays its features using holographic animations of [absurd symbols]. The crowd, made of robotic [famous figures or creatures], reacts with over-the-top applause. Add lasers, fog machines, and a giant display that says: “[ridiculous product name]”. Style: Apple Keynote meets surreal sci-fi chaos.”

We’ll be featuring the best generations in our next newsletter!

_FEEDBACK