Opus 4.5, Nvidia Boom, AWS-OpenAI Pact, EU Delay

Welcome to AI News in 10, your top AI and tech news podcast in about 10 minutes. AI tech is amazing and is changing the world fast, for example this entire podcast is curated and generated by AI using my and my kids cloned voice...

Today’s lineup moves fast: Anthropic pushes its flagship model forward with Claude Opus 4.5 and some very practical upgrades for everyday work... Nvidia’s latest earnings crush expectations and ease those 'AI bubble' jitters... OpenAI inks a seven-year, $38 billion deal with Amazon Web Services to secure a mountain of compute... the European Union hits pause on key parts of its AI Act until 2027... and Palo Alto Networks writes a $3.35 billion check for Chronosphere to fuse observability with AI-driven security. Let’s jump in.

[BEGINNING_SPONSORS]

Story one: Anthropic just rolled out Claude Opus 4.5 — with better coding, more capable agents, and smoother 'computer use' for everyday tools. Opus 4.5 is available across Anthropic’s apps and API, and it’s shipping with a 200,000 token context window plus price cuts versus earlier Opus releases. Anthropic says it’s now state-of-the-art on real-world software tasks — including an industry-leading 80.9% on the SWE-Bench Verified coding benchmark — and it’s expanding into concrete workflows like Excel modeling and a Chrome assistant. Opus 4.5 also introduces memory improvements that enable 'endless chat,' automatically distilling context so long projects don’t get interrupted. It’s live for Pro, Max, Team, and Enterprise, and developers can call it as 'claude-opus-4-5' via the API. That’s paired with new Claude for Chrome and Claude for Excel releases that broaden access beyond pilots — making the model feel less like a demo, and more like a daily driver. Company materials and reporting back up the benchmark jump and the Chrome and Excel rollout.

Why this matters: for teams that need deep reasoning but also speed and cost control, Opus 4.5 offers hybrid modes and caching to tune output quality against latency and budget. If you’ve struggled to keep long threads coherent, or to delegate multi-hour coding tasks, this upgrade targets exactly those pain points. The net effect — whether you’re building agents or doing finance models — is less hand-holding and more autonomous follow-through.

Story two: Nvidia... and the numbers are staggering. For the quarter ended October 26, 2025, Nvidia reported record revenue of $57.0 billion — up 62% year over year — with data center revenue at $51.2 billion. Gross margin hit roughly 73%, and guidance for the next quarter is an eye-popping $65 billion, plus or minus 2%. CEO Jensen Huang summed it up with a simple line: "Blackwell sales are off the charts," adding that compute demand is accelerating for both training and inference. Markets read this as confirmation that the AI spend cycle remains intact. Even without the market color, the company’s print and guidance were enough to move indices. That’s from Nvidia’s newsroom and investor releases, with further context from the financial press.

Beyond the headline, there are a few strategic nuggets. Nvidia says it set new MLPerf records, is scaling partnerships across telecom, cloud, and national initiatives, and continues to invest in networking and new data center silicon — widening the moat beyond GPUs. In other words, this isn’t just a chip story... it’s the full AI factory stack, from accelerators to interconnects to software.

Story three: OpenAI has diversified its compute footing with a seven-year, $38 billion agreement to run massive workloads on Amazon Web Services. The deal gives OpenAI access to 'hundreds of thousands' of Nvidia GPUs on AWS, with capacity coming online immediately and scaling through 2026 — with room to expand into 2027 and beyond. It’s the clearest sign yet that OpenAI intends to be multi-cloud after restructuring earlier this month — a shift that also loosened Microsoft’s exclusivity around hosting AI workloads. Amazon shares popped on the news, as analysts framed it as a vote of confidence in AWS’s AI-grade infrastructure. That’s based on wire reports and follow-on analysis of rollout timing and infrastructure details.

Why this matters: the compute bottleneck remains the gating factor for frontier models and agentic systems. Locking in assured access to advanced Nvidia clusters at this scale — on top of OpenAI’s other cloud partnerships — reduces scheduling risk for training and deployment. It also underscores how the AI era is consolidating around hyperscalers that can marshal chips, power, and networking at national-infrastructure levels.

[MIDPOINT_SPONSORS]

Story four: Europe hits pause — the European Commission will delay enforcement of the AI Act’s strictest 'high-risk' provisions until December 2027. Originally slated for August 2026, the shift follows pushback from Big Tech and is part of a broader 'Digital Omnibus' effort to streamline rules across GDPR, the e-Privacy Directive, and the Data Act. What’s being delayed affects sensitive use cases — biometric ID, health, credit, law enforcement, hiring, and road safety — while the Commission argues this is simplification, not deregulation. It’s a notable reprieve for companies building regulated AI systems, and it gives national authorities more time to stand up enforcement pipelines. Timing and scope come via Reuters reporting.

Implication: expect EU companies — and U.S. firms operating in Europe — to keep shipping features while they retool compliance programs. But it’s not a free pass... transparency and systemic-risk expectations are still rising, and the delay just shifts when the toughest checks bite.

Story five: cybersecurity meets observability. Palo Alto Networks plans to acquire Chronosphere for $3.35 billion in cash and equity. Chronosphere’s platform wrangles massive telemetry from modern apps; Palo Alto says it will plug that stream into its Cortex AgentiX so AI agents can not only flag performance issues, but also investigate and remediate them autonomously — think closing the loop from dashboard to fix. Chronosphere reported more than $160 million in ARR as of September, implying roughly 21× ARR on the price tag. The company also raised its full-year outlook alongside the deal. Details are laid out in company releases and confirmed by major business press coverage.

Strategically, this pushes Palo Alto beyond security alerts into SRE-grade reliability for AI-era workloads — where keeping clusters alive is as critical as blocking threats. It also steps squarely into a market with incumbents like Datadog and Dynatrace, suggesting more platform convergence between ops and security as AI agents take on remediation.

Quick recap before we go: Anthropic’s Claude Opus 4.5 is here, with tangible upgrades for coding, agents, and everyday tooling... Nvidia’s record quarter and $65 billion outlook solidify the AI-infrastructure boom... OpenAI’s $38 billion AWS pact locks in colossal compute through 2026 and beyond... the EU pushes 'high-risk' AI enforcement back to 2027... and Palo Alto’s $3.35 billion Chronosphere buy bets on AI agents that can diagnose and fix the stack, not just watch it. We’ll keep watching how these threads evolve — from model races to regulation to the plumbing that makes it all run.

Thanks for listening and a quick disclaimer, this podcast was generated and curated by AI using my own voice, if you want to know how I do it or want to do something similar, reach out to me at emad at ai news in 10 dot com that's ai news in one zero dot com...