☕Google Pays SpaceX $920M/Month for AI Compute

Google just paid SpaceX $920 million per month to borrow their AI compute power. That's not a typo $30 billion total because Gemini demand exploded faster than Google could build data centers. When a tech giant rents hardware from a rocket company, you know the AI race just hit a new gear.

But wait, there's more:

I'm Alex. Welcome to L8R by Innov8.

Let's dive deep 🐰

Google is running out of AI power. They just signed a massive deal to rent computer chips from SpaceX for the next few years. The search giant needs more juice to run its Gemini Enterprise AI, proving that even big tech cannot build data centers fast enough.

The details:
• Google will pay SpaceX $920 million every single month from October 2026 to June 2029.
• The deal gives Google access to about 110,000 NVIDIA GPUs located in SpaceX data centers.
• SpaceX got these massive data centers when it merged with Elon Musk's xAI in February 2026.
• Anthropic is also paying SpaceX $1.25 billion a month for similar access.
• If SpaceX fails to deliver the chips by the end of September 2026, Google can cancel the whole contract.

Why it matters:

The AI race is eating up computer chips faster than anyone can make them. Google builds some of the biggest data centers on Earth, but even they are coming up short. This deal turns SpaceX into a massive cloud provider right before its big stock market debut on June 12. It shows that whoever owns the hardware controls the future of AI.

💡 L8R's Take:

You have to admit, Google paying Elon Musk for computer chips is wild. It proves Google severely underestimated how many people would use Gemini. SpaceX is going to make an absolute fortune from this massive hardware shortage.

From Our Partner

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide

Building a smart robot just got way easier. NVIDIA released Cosmos 3 on May 31, 2026, giving developers a massive open-source AI built just for the physical world. It handles seeing, thinking, and acting all in one place instead of forcing you to use messy, separate tools.

The details:
• You can download two versions right now: a 16-billion parameter Nano model for standard GPUs and a 64-billion parameter Super model for heavy computing.
• It uses a new 'Mixture-of-Transformers' design to reason through a problem before it generates text, video, or a list of physical actions.
• NVIDIA gave it away for free under an open-source Linux Foundation license, including the code, training recipes, and datasets.
• A tiny 'Edge' version is coming soon to run directly on small robot brains in real time.

Why it matters:

Until now, making a robot smart meant tying together a bunch of different AI models. You needed one for vision, one for text, and one for movement. Cosmos 3 packs all of that into one unified system. This lets robots simulate the real world and plan safe actions without relying on locked-up APIs from rivals like OpenAI or Google.

💡 L8R's Take:

NVIDIA is playing a genius game here. By giving away the ultimate robot brain for free, they make sure every hardware startup buys their expensive chips to run it. Open-source software is great, but NVIDIA still owns the real gold mine.

Hackers love tricking AI into stealing your private data. OpenAI just dropped a massive roadblock called Lockdown Mode to stop them. It cuts off the internet so your secrets stay safe inside ChatGPT.

The details:
• Lockdown Mode blocks all outbound network requests so attackers cannot send your data to their own servers.
• You lose access to powerful tools like Agent Mode, Deep Research, and live web browsing when it is active.
• The block is hard-coded, meaning hackers cannot use a clever prompt to trick the AI into turning it off.
• OpenAI rolled this out to everyone on June 4 and 5, including Free, Pro, and Business accounts.

Why it matters:

Prompt injection attacks are a nightmare for AI security. Hackers use hidden text on websites to hijack the AI and steal your chats. By cutting the cord to the outside world, OpenAI removes the exact path hackers use to run away with your data. Companies handling sensitive info finally have a real safety net.

💡 L8R's Take:

I love this move because OpenAI finally admits AI is too easily fooled by hackers. You have to choose between cool features and real security, but that is a very fair trade. Turn this on immediately if you put private work stuff into ChatGPT.

🚀 Quick L8R Summary

We read every reply - just reply to this email and let us know how we can improve !