OpenAI's forecast? Cloudy ☁️

Plus: Google pulls back model


Welcome back! Kevin Rose has a funny rule for evaluating new AI hardware: If you feel like you should punch someone in the face for wearing it, it’s probably a bad product. It sounds like a joke, but there’s some truth in it. With AI showing up in our glasses, shirts, and social spaces, the question isn’t just what the tech can do…it’s what tech does to the vibe.

OpenAI Signs $38B Cloud Deal with Amazon

OpenAI just signed one of the largest cloud deals in tech history. Through the $38 billion agreement with Amazon, OpenAI will run and scale its AI workloads on Amazon Web Services, securing long-term GPU capacity as model sizes continue to grow.

Let’s break down what’s happening ➡️

  • In the past, OpenAI had relied heavily on Microsoft Azure for training and deployment. Now, by spreading workloads across AWS, Google, and Oracle, it reduces its risk of being tied to a single provider.

  • Bringing OpenAI workloads onto AWS reinforces Amazon as infrastructure for frontier-scale models, not just general-purpose cloud compute.

  • Training and running models at this level requires physical scale: power, chips, heat management, and long-term capacity planning. 

But scrutiny of OpenAI continues: Amid multibillion-dollar deals, OpenAI CEO Sam Altman recently pushed back on questions about OpenAI’s financial runway, saying the company is generating “well more” than $13 billion in annual revenue—but declining to elaborate.

Why it matters: This deal is about securing runway in an era of cloud turning into competitive territory. Reliable compute now determines the pace of AI progress as much as research breakthroughs do.

Google Pulls Gemma After Defamation Incident

Google withdrew one of its AI models from public access after the bot generated a fabricated sexual misconduct allegation about a US senator. The model, Gemma, produced a detailed narrative and even cited fake news links, making the accusation appear credible.

The deployment context was mismatched. Gemma was intended for developers, not for general factual Q&A. Once the model surfaced in a chatbot-like interface, users treated it as a source of truth, and the consequences followed. 

A political escalation of hallucination risk. Hallucinations are a known failure mode of language models. But when a model generates a criminal allegation involving a public official, the issue shifts from technical accuracy to reputation harm and legal liability.

Why it matters: Incidents like this create momentum for regulation that focuses on accountability for outputs, not just training methods. Is trust now the scarcest resource in AI development?

Japan Pushes Back on Sora Training Data

Japan’s top creative studios are pushing back on how their art is being used to train AI models. A consortium representing Studio Ghibli, Bandai Namco, and Square Enix has asked OpenAI to stop using their visual styles and content as training data for Sora. 

Creative differences. In the US, training is often treated as acceptable by default unless contested. In Japan, creators expect permission to be granted before training occurs. These reflect fundamentally different views of what artistic work represents. Currently there is no internationally shared standard, and models trained under one regime may face distribution limits in another.

Why it matters: The “train on everything” era is meeting its boundary: creative rights. The outcome of disputes like this will shape how AI learns from human art, and who gets a say in that process.

Turn AI into Your Income Engine

Ready to transform artificial intelligence from a buzzword into your personal revenue generator? Our groundbreaking guide, "200+ AI-Powered Income Ideas," is your gateway to financial innovation in the digital age.

Inside you'll discover:

  • A curated collection of 200+ profitable opportunities spanning content creation, e-commerce, gaming, and emerging digital markets—each vetted for real-world potential

  • Step-by-step implementation guides designed for beginners, making AI accessible regardless of your technical background

  • Cutting-edge strategies aligned with current market trends, ensuring your ventures stay ahead of the curve

Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.

Turn videos into searchable notes

Via ScreenApp

ScreenApp lets you record or upload video to automatically generate transcripts, summaries, captions, and quick answers.

How you can use it: 

  • Record meetings and get action-item summaries

  • Turn training videos into clear onboarding notes

  • Create captions for videos in multiple languages

  • Search transcripts and jump to exact timestamps

Pricing: Free and paid plans available

Build no-code custom AI assistants

Via Cubeo AI

Cubeo AI lets you create AI agents that handle sales outreach, customer support, onboarding, and internal knowledge requests.

How you can use it:

  • Send personalized sales messages at scale

  • Respond to common support questions using your documentation

  • Guide new employees through onboarding steps

  • Automate CRM updates and follow-ups

Pricing: Free and paid plans available

Jobs, announcements, and big ideas

  • Microsoft signs a $9.7 billion cloud deal with IREN as AI demand swells.

  • Walmart CEO on AI: "Every job we've got is going to change"

  • Google plans to launch AI chips into low-earth orbit, allowing them to run off solar power.

  • Amazon rolls out Fastnet, its first solo subsea cable project.

  • Apple may release its first “low-cost” Mac laptop in early 2026.

Want my honest review of ChatGPT’s new Atlas browser? I’ve got you covered.

That’s a wrap! See you for more on Friday.

—Matt (FutureTools.io)