Google Steals Spotlight from Nvidia

Nvidia just publicly congratulated a competitor. Meta is shopping for Google's chips. And ChatGPT wants to do your Black Friday shopping.

The AI landscape shifted this week in ways that have nothing to do with model intelligence. Here's what matters for your work.

Images created with Nano Banana Pro

📰 The Rundown

Google Steals the Spotlight from Nvidia—and Everyone Notices

➡️ The Move: Google's Gemini 3 launch sparked an unusual response: Nvidia tweeted congratulations while defensively noting its chips offer "greater performance, versatility, and fungibility." Meanwhile, Meta is reportedly in talks to buy Google's Tensor chips, and Anthropic announced plans to significantly expand its use of Google's TPUs.

⚡ Why This Matters: When the dominant player feels compelled to respond publicly to a competitor's product launch, pay attention. Google's custom TPUs are fundamentally different from Nvidia's GPUs—more specialized, less versatile—but that narrow focus is becoming an advantage. Salesforce CEO Marc Benioff declared he's not going back to ChatGPT after trying Gemini 3: "The leap is insane—reasoning, speed, images, video… everything is sharper and faster."

🎯 Your Takeaway: The AI arms race isn't about who builds the smartest model. It's about who controls the compute to run them at scale.

MIT's Iceberg Index Shows Where AI Disruption Is Forming

➡️ The Move: MIT and Oak Ridge National Laboratory released the Iceberg Index, a simulation tool that treats 151 million U.S. workers as individual agents, mapping more than 32,000 skills across 923 occupations in 3,000 counties. The index measures where current AI systems can already perform those skills, offering a forward-looking view of how AI may reshape the labor market down to the zip code. Read more

⚡Why this matters: This isn't doomsaying. The Iceberg Index isn't a prediction engine about exactly when or where jobs will be lost. Instead, it's meant to give a skills-centered snapshot of what today's AI systems can already do, letting policymakers explore what-if scenarios before committing real money and legislation. Tennessee has already cited the index in its official AI Workforce Action Plan, with Utah and North Carolina preparing similar reports. For individuals, this means the reskilling conversation just got a lot more concrete.

🎯 Your takeaway: The professionals who thrive won't wait for a layoff announcement. They'll treat AI fluency like a core job skill today.

Adobe Gives Creatives the Building Blocks They've Been Waiting For

➡️ The Move: Adobe officially launched Project Graph, a node-based visual editor that lets users connect Photoshop tools, AI models, and effects through an intuitive graphical interface. Think of it as building blocks for creative workflows—you design a process once, package it into a shareable tool, and reuse it across Adobe's ecosystem.

⚡ Why This Matters: This addresses the biggest frustration creatives have with AI: unpredictability. Text prompts give you randomness. Project Graph gives you control. You can fine-tune every step, share workflows with teammates, and build on community creations. The tool came from Adobe's internal incubator—a few designers saw a problem and built the solution. That origin story explains why it feels like something creatives actually asked for rather than something executives thought they needed.

🎯 Your Takeaway: The best AI tools don't replace your expertise. They let you bottle it up and reuse it.

🔧 Tool Spotlight: ChatGPT Shopping Research

What it does: Creates personalized buyer's guides by researching products across the web, asking clarifying questions, and adapting recommendations based on your feedback.

Best for: Anyone facing decision fatigue on complex purchases—electronics, home appliances, beauty products, gifts.

The key insight: This isn't instant. Shopping Research takes several minutes to compile its guide. That's intentional. OpenAI trained a specialized GPT-5 mini model to read trusted sites, synthesize information across sources, and produce actual research rather than quick summaries.

Try this: Ask it to help you find a gift for someone specific: "I need a gift for my four-year-old niece who loves art." It will ask about your budget, her specific interests, and whether you want something educational or purely fun. The result is a curated guide, not a generic list.

Access: Available now to all logged-in ChatGPT users on Free, Go, Plus, and Pro plans. Select "Shopping Research" from the (+) menu or let ChatGPT suggest it automatically when you ask a shopping question.

👉 Try This Today: The Comparison Table Prompt

Time required: 5 minutes

Before making any significant decision this week—software tool, vendor, hire, strategy—try this prompt structure:

I need to decide between [Option A], [Option B], and [Option C].

My priorities are:
1. [Most important factor]
2. [Second priority]
3. [Third priority]

My constraints are:
- [Budget/time/resource limit]
- [Non-negotiable requirement]

Ask me clarifying questions to make sure you understand my priorities and constraints.

Create a comparison table scoring each option 1-5 on my priorities. 
Then recommend which option best fits my specific situation and explain why.

Why it works: Forcing yourself to articulate priorities and constraints before asking for comparison changes what you get back. Allowing the AI to ask questions makes sure it understands what you need. You're not asking "which is best?" You're asking "Let's have a discussion to discover which is best for me?"

✨ The Wire

⊡ ChatGPT hits 800 million weekly active users, doubling from 400 million in February. Sam Altman says growth is accelerating as AI becomes infrastructure rather than novelty. TechCrunch

⧖ Gemini 3 sits atop benchmark leaderboards for text generation, image editing, and image processing, putting it ahead of ChatGPT, Grok, and Claude in those categories. Different models still lead in search and specialized tasks. CNN

⌘ DeepSeek open-sources Math V2, a Chinese model that won gold at IMO 2025 and scored 118/120 on Putnam 2024. Training cost: $294K—roughly 1000x cheaper than Western competitors. NinjaAI

⟡ OpenAI's shopping accuracy: 52% vs 37% for standard ChatGPT Search on complex product queries with multiple constraints. Still far from perfect, but meaningful improvement for decision-heavy research. Search Engine Journal

📚 Go Deeper

The real story behind Google's chip ambitions: CNN breaks down how Google's custom TPUs are attracting competitors like Meta and Anthropic, and what that means for Nvidia's dominance. Required reading if you want to understand the infrastructure layer beneath the AI hype. Read →

Everything happening with ChatGPT in 2025: TechCrunch maintains a comprehensive timeline of every ChatGPT update this year. Useful reference when you're trying to remember what features are actually available now versus announced but not shipped. Read →

Go make something happen.

Neural Notes — AI that amplifies your value, not replaces it.