đź’°OpenAI Just Dropped $300B on Compute

PLUS: How to create fully explorable 3D worlds

Author

Alvaro Cintas
September 11, 2025

Good morning, AI enthusiasts. OpenAI just signed one of the largest cloud deals in history, a $300 billion contract with Oracle to fuel its next wave of AI growth.

PLUS: How to create fully explorable 3D worlds

TODAY IN AI

OpenAI has reportedly struck one of the largest cloud computing deals ever, spending $300 billion over five years to secure 4.5 gigawatts of computing power from Oracle, starting in 2027. The contract is part of Project Stargate, a mega data center initiative announced with SoftBank and former President Trump.

→ $300B deal spans 2027–2032
→ 4.5 GW data centers to power OpenAI’s AI expansion
→ Oracle’s cloud revenue up 77% YoY, stock surges
→ Larry Ellison becomes world’s richest person

Why is important: This deal shows the sheer scale of AI’s compute hunger, with OpenAI locking in resources years in advance. It also cements Oracle as a top-tier AI infrastructure player, challenging AWS, Azure, and Google Cloud in the race to power frontier models.

AI coding platform Replit has secured $250 million in fresh funding, tripling its valuation to $3 billion. The round was led by Prysm Capital, with Google’s AI Futures Fund and Amex Ventures joining as strategic investors. Existing backers like a16z and Coatue also doubled down.

→ Valuation jumped from $1B (2023) → $3B (2025)
→ Annualized revenue surged from $2.8M → $150M in under a year
→ Customers include Duolingo & Zillow

Why is important: The funding highlights the surging demand for AI-assisted coding tools. With competitors like Cursor ($10B valuation) and Cognition ($10.2B valuation), Replit’s push into enterprise “vibe-coding” for non-technical teams may be its biggest differentiator.

HOW TO AI

How to create fully explorable 3D worlds

HunyuanWorld-Voyager is China’s breakthrough 3D world model that can transform any text, image, or video into a fully explorable 3D environment in just minutes. The best part is that you can step inside your worlds instantly and use them in VR, games, or simulations for completely free.

🧰 Who is this for

  • Game developers creating immersive environments

  • VR/AR designers building interactive experiences

  • Researchers and educators needing 3D simulations

  • Creators and hobbyists curious about world-building with AI

STEP 1: Access the Tool

Go to the official HunyuanWorld website. On the main dashboard, you’ll see a clean interface with a prompt field at the bottom and options to upload an image on the side.

If you want to start with just text, you can describe the idea of the world you want to create in the prompt field.

For example:
"A vast alien desert with glowing crystal towers and floating bridges under two suns."

Within minutes, your first 3D world will be ready to explore.

STEP 2: Generate with Image

For more accurate results, it’s best to start with an image rather than relying only on text.

Click the image icon next to the prompt field and select the image you want to use. Once your image is uploaded, click the Generate Instantly button on the right-hand side.

STEP 3: Explore and Control Your World

Once the generation is complete, you’ll see two options at the top of your screen: 360 Panorama and Roaming Scene.

360 Panorama gives you a full spherical view of your generated world. You can look around in every direction, almost like being inside a giant immersive photo. This is perfect for quick visualization, concept design, or background environments.

Roaming Scenes lets you move inside the world, not just look around. Use W, A, S, D to walk and your mouse to look in any direction. You can explore landscapes, enter structures, and see details up close, just like being inside a real game.

ESSENTIAL BITES

  • ByteDance introduced Seedream 4.0, a new image generation and editing model that competes with Nano Banana with 4K outputs and multimodal capabilities.

  • China successfully tested an AI-enhanced radar system that maintained 99% target tracking accuracy under heavy electronic warfare jamming, compared to 70–80% for traditional systems.

  • Amazon is working on new AR glasses codenamed Jayhawk, featuring a camera, speakers, and a display, with a launch expected in 2026–27. A bulkier version called Amelia, designed for delivery drivers, could see 100,000 units deployed in 2026.

  • Boston startup and MIT Media Lab spinoff Alterego unveiled a “near-telepathic” wearable that uses AI to detect subtle movements and interpret a user’s silent communication “at the speed of thought,” no speaking required.

⚙️ Hot AI Tools:

That's it! See you tomorrow

- Dr. Alvaro Cintas