💸 Anthropic is tired of paying the Nvidia tax

Plus, Google's Top AI Brain Just Left

August 05, 2026

In partnership with

Welcome back, AI Admirers!

Breaking News: Anthropic, the company behind the popular AI assistant Claude, is building a new engineering team to design custom, in-house AI chips to optimize performance and reduce reliance on third-party hardware. 

Get ready to dive into the latest happenings in AI.

📢 Today's Headline:

  • Anthropic Is Building Custom AI Chips

  • Google Chief Scientist Quits

  • The $1.2 Trillion Bet on Orbital AI Chips

  • Google Assistant Is Shutting Down

  • Latest AI Tools & Resources

  • Today’s Poll and Results

Read time: 4.0 minutes!

22 plug-and-play ChatGPT agents

22 ChatGPT Agents Built for Every Marketing Job

Most marketers use ChatGPT to do general research and then call it an AI strategy. The ones outperforming them are deploying specialized agents built for specific jobs.

We put together 22 plug-and-play ChatGPT marketing agents that handle the work eating your week, each with built-in instructions and structured outputs ready to go in under 5 minutes.

Subscribe to Marketing Against the Grain and get all 22 free.

Inside you'll find:

  • Competitive intelligence agent that visits competitor websites and builds detailed comparison matrices automatically

  • Customer feedback analyzer that ranks improvement opportunities by business impact

  • Social listening specialist that monitors brand mentions and flags reputation risks before they escalate

  • Campaign optimization agents that handle attribution analysis and surface what is actually driving results

Your competitors are already running agents like these.

Get 22 ChatGPT Marketing Agents free when you subscribe to Marketing Against the Grain today.

Hand-picked News

Anthropic is assembling an in-house team to build custom chips for Claude, signaling that relying entirely on Nvidia and standard cloud providers isn't sustainable anymore.

  • Inference over training: This silicon focuses specifically on running Claude day-to-day, aiming to make user queries faster and drastically cheaper per token.

  • Samsung partnership: Rumors point to Samsung for manufacturing, giving Anthropic a dedicated production line while competitors scramble over TSMC’s limited capacity.

  • Urgent timeline: Job listings demand engineers who can ship semiconductor designs fast, proving this is a priority product push rather than an R&D project.

AI labs are learning that controlling the algorithm isn't enough when hardware access dictates who survives. Custom silicon is fast becoming the new baseline requirement for top-tier AI labs trying to keep margins healthy.

👀 The takeaway: Custom hardware will mean cheaper API rates and lower latency for Claude users. If you build on top of LLMs, track these hardware shifts as they will directly impact your operational costs.

💡 Try this now: Benchmark your app's current response times and token costs on Claude 3.5 Sonnet now so you have a baseline to compare against as dedicated hardware rolls out.

Jeff Dean and three top Google AI leaders left after nearly three decades to launch Discovery Loop, a startup focused on fully automating the scientific method.

  • The architects left: Dean is joined by Sanjay Ghemawat, Oriol Vinyals, and Quoc Le—the engineers behind MapReduce, TensorFlow, and core Gemini features.

  • Self-driving research: Discovery Loop builds AI agents that form hypotheses, execute virtual experiments, analyze data, and refine results autonomously.

  • Google funded its own departure: Sundar Pichai couldn't keep the team inside Google, so Alphabet joined as a founding investor and cloud compute provider.

When the architects of modern computing pivot to autonomous scientific loops, it signals that the next frontier isn't better chatbots, but compressing years of physical R&D into days.

👀 The takeaway: Watch AI-driven research platforms closely. Agentic loops will soon drive massive breakthroughs in biotech, materials science, and energy by eliminating human bottlenecks in testing.

💡 Try this now: Experiment with multi-agent frameworks like AutoGen or CrewAI to identify where iterative loops can automate multi-step research or analysis tasks in your workflow.

SpaceX is asking regulators for permission to launch up to one million satellite data centers into orbit, proving that Earth's power grid is failing to keep up with AI compute demand.

  • One million AI satellites: The filing requests permission for more satellites than every man-made object currently orbiting Earth combined, all packed with radiation-hardened AI processors.

  • Solar power without night: In high orbit, these servers can harvest continuous, unshielded sunlight without taking up land or consuming trillions of gallons of cooling water.

  • The heat problem moves to vacuum: Without air to conduct thermal energy, keeping high-powered GPUs from melting in space requires massive, unproven radiator systems.

Ground-based data centers are triggering local power grid panics and water shortages, forcing tech giants to buy up entire nuclear plants just to keep pace with model training demands.

⚠️ Power grid limitations are now the single biggest bottleneck in tech history. Whether orbital compute actually succeeds or gets crushed by cosmic radiation, top AI operators are already looking beyond planet Earth to avoid hitting a total compute wall.

💡 Try this now: Audit your team's API usage for token efficiency today, because as compute capacity gets tighter and more expensive, trimming prompt bloat will directly protect your margins.

Quick Hits

🔥 New Tools & Resources

🔥 NextDoor.Company - Discover startups hiring near you, on a map

🔥 Wispr Flow Notetaker - Meeting notes that get the details right.

🔥 Gamma - Create unlimited presentations, websites, and more in seconds

🔥 Unlimited Prompts - Get 10k+ ChatGPT prompts now

🔥 AdAnt AI - Claude for viral, high-converting social ads

From our Partner

Know Exactly Who's Spending Your AI Budget.

Every AI request leaves a trail. Mesh gives engineering and finance complete visibility into who used which model, how many tokens were consumed and where your AI budget is going.

Stop guessing. Start governing AI spend.

Connect once, switch between GPT, Claude, Gemini and hundreds more whenever you want, while automatically routing requests for 40% lower costs and 99.99% AI response rate.

🚨 Quick Poll

Today’s Poll:

Should Anthropic just go all-in on its own chips instead of still buying from Nvidia, AWS, Google, and AMD?

Login or Subscribe to participate in polls.

Vote today, see the results tomorrow!

Previous Poll:

Do you believe OpenAI's claim that it doesn't have or want Apple's trade secrets?

  • A) Yes – I trust OpenAI's statement – 10%

  • B) No – Where there's smoke, there's fire - 90% 🏆

People aren't buying corporate denials at face value anymore — especially in a space this cutthroat. When the default assumption is "guilty until proven innocent," it says less about OpenAI specifically and more about how little trust remains in Big Tech's self-policing.

🚀 The Only AI You Ever Need

Your ideas deserve to be seen. Gamma helps you turn them into presentations, websites, and content in minutes — no design bottlenecks, no blank page stress.

📌 Key Features

  • Presentations: Build polished decks, export to PPT or PDF.

  • Documents: Create one‑pagers, reports, or white papers with structure.

  • API: Automate and scale content creation across your tools.

  • Social Media: Generate posts sized and styled for each platform.

  • Websites: Launch a hosted site in minutes — no coding needed.

  • Graphics: Design custom visuals, from infographics to illustrations.

Skip the blank page. With Gamma, you can create, edit, and share ideas faster than ever — from decks to websites to social posts.