OpenAI & AWS Power Play

Hey folks,

Here’s a fun one to kick off the mid-week. Sam Altman went on X to show that he tried to cancel his Tesla Roadster which he reserved in 2018, and now 7.5 years later asked for his $50k Roadster deposit to be refunded after showing a bounced email with "Address not found" when trying to cancel. Elon spat back a series of tweets and refunded his $50k within 24hrs.

The two co-founded OpenAI in 2018, and Musk's been accusing Altman of turning the nonprofit into a commercial enterprise ever since. The latest public spat over cancelling a Tesla just ads more fuel to the fire, and we’re all here with 🍿 watching the AI Chiefs battle it out.

Let’s dive in..

☁️ OpenAI Signs $38B AWS Deal, Breaks Microsoft Exclusivity

🐝 The Buzz: OpenAI announced a seven-year, $38 billion agreement with AWS, securing immediate access to hundreds of thousands of Nvidia GB200 and GB300 GPUs clustered via Amazon EC2 UltraServers for low-latency distributed training. The deal marks OpenAI's largest infrastructure commitment and its most significant diversification from Microsoft since exclusive cloud provider terms expired under newly negotiated commercial agreements. Amazon stock closed 4% higher at a record high following the announcement.

• Multi-cloud infrastructure strategy: OpenAI explicitly reducing Microsoft dependency after years as exclusive cloud provider.

• Immediate GPU access: Unlike typical capacity commitments requiring buildout, OpenAI starts using AWS compute now, addressing capacity bottlenecks that have slowed model iteration.

• EC2 UltraServers clustering: Hundreds of thousands of Nvidia GPUs connected on same network fabric for low-latency communication critical to distributed training of large models like GPT-5 and o1.

💡 Takeaway: OpenAI's Microsoft exclusivity ending validates multi-cloud AI infrastructure strategies for enterprises. Vendor lock-in for AI workloads is no longer strategic and expect frontier labs and large engineering orgs to negotiate similar flexibility. For technical leaders, this signals you can push back on single-vendor commitments for AI compute.

Together with Attio

Introducing the first AI-native CRM

Connect your email, and you’ll instantly get a CRM with enriched customer insights and a platform that grows with your business.

With AI at the core, Attio lets you:

• Prospect and route leads with research agents

• Get real-time insights during customer calls

• Build powerful automations for your complex workflows

Join industry leaders like Granola, Taskrabbit, Flatfile and more.

👉Try Attio Pro for free

📊 74% of Businesses Report Positive AI ROI, Study Finds

🐝 The Buzz: A Wharton School study found that 74% of businesses measuring generative AI ROI are seeing positive returns. The research contrasts sharply with a previous MIT study claiming 95% of AI pilots were failing, using a broader ROI definition that includes individual productivity, employee retention, and operational efficiency, not just revenue impact. Tech/telecom leads at 88% positive ROI, banking/finance at 83%, while retail lags at 54%.

• ROI measurement broadening: Study moves beyond revenue metrics to include productivity gains, employee retention improvements, and operational efficiency and signals shift from pure financial returns to holistic business impact.

• Industry variance: Tech/telecom seeing 88% vs retail's 54% suggests infrastructure and digital maturity matter more than AI technology itself for successful deployment.

• Leadership perception gap: 12-point gap between VPs (81%) and managers/directors (69%) indicates either executive optimism disconnect or middle management implementation reality check.

• Smaller companies adapting faster: Study shows nimbler organizations deploying AI more successfully than large enterprises, likely due to fewer legacy systems and faster decision cycles.

💡 Takeaway: Self-reported AI ROI hitting 74% challenges the "AI pilots are failing" narrative, but the broader ROI definition matters. For engineering leaders, this suggests measuring productivity and operational efficiency instead of pure revenue may be more realistic for AI value assessment.

📰 Like newsletters? Here are more our readers enjoy.

💻️ How to use ChatGPT Atlas Automations

🐝 AI buzz bits

🎬 Coca-Cola released its second AI-generated holiday ad, and creatives are still mad. Coke exec says "the craftsmanship is ten times better" than last year. Online reactions are still very mixed. Turns out people actually liked the human-made magic.

🌐 ChatGPT Atlas browser launched agent mode for paid users. It'll book reservations, order groceries, and fill out forms for you. Finally, a browser that does your chores. Free users get the base browser without the robot butler.

🛠️ Chrome DevTools 142 added Gemini code suggestions and Debug with AI for performance traces. AI moved from external copilots straight into your browser's native dev tools. Now your debugger can explain why your code is slow.

🔥 AI productivity tools

For a full list of 1000+ AI tools, visit our Supertool Directory

🤘 Midweek Madness