OpenAI Signs $38B Cloud Deal With Amazon to Chase Infinite Compute

Welcome back! The money behind AI is swelling to extraordinary levels. OpenAI just locked in a cloud deal with Amazon, showing that the future of intelligence depends on who controls the servers. Companies are counting real returns on their AI investments, creative industries are experimenting with machine-made art, and startups are racing to legitimize their data sources. The gold rush for compute power has begun, and everyone wants a claim.

In today’s Generative AI Newsletter:
• OpenAI’s $38B cloud deal with Amazon
• Wharton finds companies are finally seeing ROI from AI
• LG heir funds a new AI-powered film venture in Asia
• Perplexity licenses Getty Images to clean up its data footprint

OpenAI has struck a $38 billion deal with Amazon to buy cloud computing power over the next seven years, marking one of the largest infrastructure commitments in AI history. The company said it will begin using AWS compute immediately, with full capacity expected by the end of 2026 and room to expand beyond 2027.

Here’s what’s happening:

The deal follows OpenAI’s internal restructuring last week, which lifted a previous requirement to get Microsoft’s approval for cloud partnerships. Some analysts see a bubble forming as billions pour into this technology. If compute is the new oil, OpenAI just bought a refinery the size of a continent and the market is still deciding whether it runs on fuel or fumes.

Special highlight from our network

This weekend, join Outskill’s live 2-day AI Mastermind and learn how to build GPT agents, automate 80 percent of your workflow, and apply AI across content, marketing, and business operations. These are the same systems used by Fortune 500 teams, and now you can learn them without code, without fluff, and without paying $395.

Part of the Black Friday offer. First 500 seats are free.
Live sessions run Saturday and Sunday, 10 AM to 7 PM EST.

👉 Reserve your spot now before it fills up

A new Wharton Human–AI Research report shows that enterprise AI has transformed. After two years of trial runs and cautious pilots, most large U.S. companies now treat GenAI as a real part of their workflow. The study, based on 800 senior decision-makers, found that 82% use AI weekly, and three out of four are already tracking measurable ROI in productivity and profit.

Here’s what the data shows:

The study shows a clear turn toward “accountable acceleration” and the hype cycle is fading slowly. Enterprises are moving past asking if AI works. Instead, they’re asking how to make it work better. For an industry accused of chasing buzzwords, that’s the most radical outcome of all.

Brian Koo, grandson of LG’s founder, is funding Utopai East, a new venture between Stock Farm Road (SFR) and Utopai Studios that wants to bring AI into film production. The company plans to use AI for tasks like editing, scheduling, and visual planning, while keeping creative control with writers, directors, and actors.

Here’s what’s in motion:

Filmmakers remain cautious, with many still uneasy about AI’s role in storytelling. Yet others see tools like Utopai’s as a way to cut costs and open access to global IP. The truth may land somewhere in between. The art could still belong to humans, but the process of making it might soon speak a new language.

AI search startup Perplexity has signed a multi-year licensing agreement with Getty Images, giving it the right to display Getty’s photos across its AI tools. The deal comes after repeated accusations that Perplexity scraped content without permission, including a lawsuit from Reddit earlier this year. By adding Getty’s imagery, Perplexity is trying to move from gray-area data collection toward more legitimate partnerships.

Here’s what’s changing:

The agreement is a small but meaningful correction for a company often accused of cutting corners on data sourcing. It’s both a legal safeguard and a credibility play at a time when copyright scrutiny around AI remains intense. Yet as Stanford’s Mark Lemley points out, signing licenses while defending fair-use claims could weaken the very arguments these companies rely on in court.