In partnership with
🌟 Editor's Note: Recapping the AI landscape from 02/03/26 - 02/09/26.
🎇✅ Welcoming Thoughts
Welcome to the 30th edition of NoahonAI.
What’s included: company moves, a weekly winner, AI industry impacts, practical use cases, and more.
Wow, lots of Superbowl AI commercials.
I thought the Claude ones were pretty good.
Seems lots of people still look at different AI models and see the same thing.
Interview today with someone I met at a UChicago PhD event in late 2025 (I was there with a friend not as a student).
GPT Voice mode somehow manages to get worse every week.
Anthropic released new tools and the market went into shock.
Was surprised to see an AOL article in NVIDIA5 research.
Interesting article on building a $20,000 C Compiler with Claude Code.
Alphabet (Google) crushed earnings. No surprise there.
Not a ton of action this week across the board.
I wonder if physical AI objects and wearables will become more popular once OpenAI launches theirs.
Creating videos from the Terminal in seconds is pretty cool (See PUC).
Let’s get started—plenty to cover this week.
👑 This Week’s Winner: NVIDIA
NVIDIA is back on top this week with a strong performance spanning industrial software, global infrastructure, and capital markets. As AI CapEx continues to rise, NVIDIA is showing they are not going anywhere. Here’s the details:
NVIDIA x Dassault Partnership: Joined to build “industry world models” that combine NVIDIA’s AI compute with Dassault’s simulation software. The goal is highly accurate, science-backed digital replicas of real-world systems. This is very cool. Mastering simulations at scale will lead to valuable new data.
Chips Power xAI: Apollo is providing $3.4B financing for xAI centered on leasing NVIDIA GPUs, underscoring how NVIDIA hardware remains the default foundation for frontier AI training. When in doubt, spend billions on NVIDIA chips.
Finalized $380M Taipei Deal: A royalty agreement tied to its planned Taiwan headquarters, removing a key hurdle and doubling down on Taiwan as a core base for engineering. A critical node in the global AI hardware supply chain. Brings NVIDIA physically closer to the factories they rely on.
More From NVIDIA: In the background, reports suggest their long-rumored OpenAI investment is back on track at roughly $20B. At the same time, NVIDIA has reportedly delayed new RTX gaming GPU refreshes as AI demand and memory constraints pull resources toward higher-margin data center chips. NVIDIA is prioritizing AI infrastructure, partnerships, and long-term platform control, even if that means sidelining legacy markets.
From Top to Bottom: Open AI, Google Gemini, xAI, Meta AI, Anthropic, NVIDIA.
⬇️ The Rest of the Field
Who’s moving, who’s stalling, and who’s climbing: Ordered by production this week.
🟣 Google // Gemini
Earnings Success: Alphabet posted ~$113.8B in Q4 revenue and ~$34.5B profit, beating expectations, with strength in Search and Cloud and AI cited as a growth driver. No surprise here. Extremely bullish on Google!
Liberty Global AI Partnership: Google Cloud signed a 5-year deal with Liberty Global to embed Gemini into products like Horizon TV and expand AI for network ops, security, and customer service across Europe. Nice move.
2026 AI Capex Surge: Alphabet said it plans to nearly double 2026 capex to ~$175B–$185B (vs. ~$92B prior year), largely to build more AI infrastructure like data centers and TPU/GPU capacity. Common theme across the NVIDIA5. AI spend is booming.
🟢 OpenAI // ChatGPT
Growth is Back: CEO Sam Altman told staff ChatGPT is back to 10%+ monthly user growth, and Codex usage jumped ~50% WoW after the macOS Codex app launched. Seeing a lot of positives with OpenAI right now. Coding (Codex) improvements seem to be real.
GPT Ads Are Here: OpenAI started U.S. only ad testing for logged-in adults on Free and Go, showing clearly labeled sponsored units below answers, excluding sensitive topics and minors. Honestly this makes sense for GPT’s target market compared to a Gemini or Anthropic. Curious to what public reaction will be. I think people will get used to it or pay for Pro.
Frontier Enterprise Platform: OpenAI launched Frontier, an enterprise system to build and govern AI agents across internal tools. Early adopters including Intuit, Uber, State Farm, HP, Thermo Fisher, and Oracle. Impressive list. Open AI is going after Anthropic’s two strongest domains (Coding + Enterprise). I’m seeing huge demand for agents right now.
🟠 Anthropic // Claude
Claude Opus 4.6 launch: Anthropic released Claude Opus 4.6 with a ~1M token context window, “agent teams” that work in parallel, and stronger reasoning and coding, positioning it for long-document and enterprise-scale workflows. The best AI model just got a bit better.
Goldman Deploys Claude Agents: Goldman Sachs confirmed using Claude-powered autonomous agents to automate internal workflows such as trade accounting, transactions, due diligence, and onboarding. Cool! Intrigued by the latest Anthropic push into financial tools. Will test soon so I can recommend.
Super Bowl vs. OpenAI: Anthropic ran Super Bowl ads poking at OpenAI’s move toward ads, leaning into Claude’s ad-free positioning; the spot drew public criticism from OpenAI leadership. I thought these were pretty good! Reception has not been great.
🔵 Meta // Meta AI
Super Bowl AI Glasses: Meta ran two Super Bowl ads pushing its AI smart glasses (Oakley Meta), using big-name creators/celebs to signal a major consumer hardware + AI wearables bet. Seems like they’re pushing these hard. I haven’t seen a physical consumer AI product that I’m a fan of yet.
EU targets WhatsApp: EU antitrust regulators accused Meta of favoring Meta AI by restricting rival chatbots on WhatsApp’s Business API, and warned they could impose “interim measures” while the case plays out. Seems like EU is much quicker to legislation than U.S. - makes sense here.
Sacramento State AI: Meta/Zuck pledged $50M toward a new downtown Sacramento State campus project that includes an AI center aimed at building local STEM/AI talent and revitalizing the area. Big fan of investing in AI tools/growth in universities.
🔴 xAI // Grok
Apollo Lines Up $3.4B Chip-Financing: Apollo is nearing a deal that effectively funds NVIDIA GPU purchases via a lease/financing structure for xAI, helping it scale compute for Grok and other model workloads without paying all upfront. Saves cash while slightly decreasing risk. Smart.
UK Opens Probe Into xAI: The UK Information Commissioner’s Office launched a formal investigation into xAI tied to how Grok processes personal data and related risks around harmful imagery. xAI still feeling blowback from failure to get Grok images under control.
SpaceX Prioritizing a Moon City: Musk said SpaceX has shifted near-term focus toward building a self-growing city on the Moon (targeting <10 years), citing faster iteration vs. Mars timelines. Curious to see what role newly acquired xAI plays in SpaceX long term goals.
🎨 Impact Industries 🚑
Creative Industries // Amazon Films
Amazon plans to begin testing AI tools for film and TV production next month, bringing generative tech directly into the studio workflow. The pilot is aimed at speeding up parts of development and production, like early concepting, iteration, and other repetitive tasks that typically slow teams down. If the tests go well, it signals a big shift: AI stops being a side experiment and becomes part of how shows get made at scale today.
Read the Story
Healthcare // AI + 3D Custom jaw
After 17-year-old Mya Buie’s jaw was shattered by a gunshot, doctors in Iowa used AI to turn scans of her face into a precise surgical plan. They sent the digital model to a lab, where a 3D printer produced a custom jaw plate sized to her anatomy. The surgeon said it’s far easier than hand-bending titanium, and Buie is expected to fully recover in time to restore her bite, appearance, and four missing teeth. The value of AI in this case is speed and efficiency.
Read the Story
🎙 Weekly Interview: 10 Minutes With Ruby Ostrow
Ruby Ostrow
🏠 Background: Ruby holds a Bachelor’s in Computer Science and Classical Studies from Bard College and an MSc in Artificial Intelligence from the University of Edinburgh. Her background blends high-level technical NLP research with a humanities-driven approach to ethics and AI communication.
💼 Work: Ruby is an AI Evaluation specialist at Nuarca Labs, a startup building RAG-based LLMs for the financial sector. She focuses on developing frameworks to verify the accuracy of automated regulatory compliance and document generation in highly conservative, high-stakes environments.
🚀 Quote: “Large language models are only one very small sector of what AI actually means.”
🎙️ Condensed Interview Transcript — Ruby Ostrow
Question 1
Noah: What is your general take on the current state of the AI space?
Ruby: It’s an interesting intersection of rapidly changing tech and shifting public opinion. We’re seeing a lot of market oversaturation with competing models and tools. Because people don’t fully know how to use them yet, they see it's not working perfectly and wonder why so much money is being pumped into it. We’ve reached a point where model performance is continually fluctuating.
Question 2
Noah: What does your day-to-day tech stack look like for AI?
Ruby: I use ChatGPT frequently and have used Copilot in VS Code for debugging, though I’ve been putting out feelers for new tools lately. I’ve been experimenting with Gemini and Codex, but over the past couple of weeks, I’ve leaned heavily into Claude. Its performance in debugging complex code has been very impressive compared to other models.
Question 3
Noah: What do you learn in a dedicated Master’s in Artificial Intelligence program?
Ruby: It’s a challenge because the field changes every week. My program focused on machine learning to show the breadth of AI beyond just LLMs. The NLP courses had to be completely restructured to keep up with the explosion of developments. It’s about learning the history, the machine learning foundations, and the ethics of how these systems operate.
Question 4
Noah: What are you most excited about regarding the 5 to 10 year future of AI?
Ruby: I’m particularly interested in how AI can increase accessibility. It can assist people entering new careers, whether in coding or finance, by acting as a bridge for those coming from different educational backgrounds. If we focus on the areas where AI is actually helpful, it has the potential to do a lot of good for people.
Question 5
Noah: What are your primary concerns regarding public perception and AI integration?
Ruby: My main concern is AI being used for high-stakes decisions on individuals, like resume reviews or school integrations, without any human review. There’s also the concern of sensitive data exposure. I think we need to regain some perspective on how these models actually operate and bring the conversation back down to earth regarding the level we're actually at.
👨💻 Practical Use Case: Remotion In Claude Code
Difficulty: Advanced
Remotion is a developer-first way to create videos using code instead of timelines. Rather than dragging clips around in a traditional video editor, you define layouts, animations, and timing in JavaScript/React, then render the final video programmatically. This makes video creation dynamic, repeatable, and scalable, especially when paired with an AI coding assistant like Claude Code.
I like this because it fits into a normal coding workflow. You can stay in your terminal, describe what you want, and let Claude Code help scaffold, modify, or debug the video logic without opening a traditional editor.
I told Claude Code to go to my website and create a promotional video. It took about 30 seconds, and I did it while I was already in the terminal on my Mac. No timeline editing, just code and a rendered output. Here’s how it turned out 👇
This is where Remotion really shines:
Generating personalized videos at scale from structured data.
Producing templated short-form clips programmatically for demos.
Updating branding, layouts, or messaging across an entire video library.
Claude Code is useful here because it can help you quickly get from idea → working render, set up project structure, and adjust timing/animations, all from a single place, while multitasking.
Learn more below ⬇️
🏥 Startup Spotlight
OpenEvidence
OpenEvidence — The AI Powered Brain for Doctors.
The Problem: Medical knowledge is currently doubling every 73 days. For a doctor to stay truly up-to-date by reading just the top journals and guidelines, they would need to spend roughly nine hours every single day just reading. This "information firehose" leads to burnout and, more dangerously, a gap between the latest life-saving research and actual patient care.
The Solution: OpenEvidence is a specialized AI search engine that acts as a clinical co-pilot. Unlike general LLMs that might hallucinate, OpenEvidence only draws from 35 million peer-reviewed medical publications. Every answer it provides is grounded in direct, hyperlinked citations, allowing physicians to verify facts instantly at the point of care. If the evidence isn't there, the AI simply won't answer.
The Backstory: The company was founded in 2021 by Daniel Nadler and Zachary Ziegler. Nadler is a "renaissance man" founder—a Harvard PhD who previously sold his financial AI startup, Kensho, for $550M, and is also a published poet. Since launching, OpenEvidence has seen the fastest adoption of any medical tech in history; it is now used by over 40% of all verified physicians in the U.S.
My Thoughts: I think the numbers speak for themselves here. It’s similar to NotebookLM or any RAG system in that all of the data is pulled from one place without leaving it open to outside interpretation. In fields like medical and legal, where hallucination can cost lives, that’s incredibly important. The scale of data is also very impressive. I can see this being very useful, and a good, practical introduction to AI usage in potentially resistant industries. Good stuff.
“It’s not likely you’ll lose a job to AI. You’re going to lose the job to somebody who uses AI”
- Jensen Huang | NVIDIA CEO
Did you have a favorite Superbowl AI ad? A least favorite? Till Next Time,
Noah on AI