• Lore Brief
  • Posts
  • Issue #151: NVIDIA Bets on Intel to Rewire Compute

Issue #151: NVIDIA Bets on Intel to Rewire Compute

PLUS: Luma’s Ray3 video model, Meta’s AI glasses, and more signals shaping the future

Author

Nathan Lands
September 19, 2025

Good morning, welcome to this week’s Lore Brief, your 3-minute brief of the most important moves in AI and tech.

This issue is brought to you by Factory, the fastest way to ship software with autonomous engineering agents.

P.S. Subscribing to The Next Wave helps us push further and land the guests you want to hear from.

  • NVIDIA bets $5B on Intel → The two longtime rivals just formed a chip alliance. NVIDIA GPUs + Intel CPUs = a re-wiring of the compute stack. This signals how scarce and strategic compute has become. Read more →

  • Luma launches Ray3 video model → The world’s first “reasoning video model” promises cinematic HDR video from text prompts. This could be the tipping point where AI video goes mainstream for creators. Read more →

  • Meta unveils Ray-Ban Display glasses → Zuckerberg showed off AI glasses with gesture controls, in-lens displays, and real-time subtitles. Still glitchy, but a step toward wearable AI platforms. Read more →

Vibe coding with GPT-5 Codex Is Getting Insanely Good

ElevenLabs gets even better

Great example of the power of Nano Banana

Universal memory API & app to retain, organize, and retrieve your “knowledge state” across AI clients.

  • Best for: Folks building AI agents, personal second-brains, or anyone wanting context persistence across tools like ChatGPT, Claude, Cursor.

  • Skip if: Your workflow is simple and you don’t need long-term memory or cross-tool integration.

  • Gotcha: Depending on scale, memory storage + inference can add latency or cost; prompt context limits may still bite in edge-cases.

OpenAI just published an in-depth, 63-page study detailing how hundreds of millions of people are using ChatGPT. For entrepreneurs and anyone building AI-powered products and apps, this report is a fascinating look into user behavior, emerging trends, and the real-world value of generative AI. (Link to the pdf of the full study on the OpenAI website.)

The main take aways:

1. Personal use unsurprisingly dominates as more and more people start using ChatGPT

  • Non-work-related messages have grown from 53% to over 70% of all consumer usage.

  • While work-related use is growing, non-work queries are increasing at an even faster rate.

2. Moving from mostly male users to gender parity

  • Initially, about 80% of active users had typically masculine first names.

  • That gap has closed dramatically over time, reaching near-parity in early 2025.

  • As of June 2025, active users are now slightly more likely to have typically feminine names, indicating a significant broadening of the user base.

3. Users are "Asking" more than "Doing"

  • The study categorizes intent into "Asking" (seeking information/advice), "Doing" (requesting a task), and "Expressing".

  • "Asking" is the most common intent at 49% of all messages and has grown faster over the last year than "Doing" (40%).

  • Users also rate "Asking" interactions as being of higher quality, suggesting people find significant value in using AI as a decision-support tool or "co-pilot".

4. At work, it's more about polishing than creating

  • "Writing" is the most common work-related activity, making up 40% of work messages.

  • Interestingly, two-thirds of these "Writing" requests involve modifying user-provided text - such as editing, critiquing, or translating - rather than generating new content from scratch.

  • This points to a strong need for AI tools that help professionals refine and improve their own communications.

5. Coding and Companionship Are Not the Mainstream (Yet)

  • Despite significant media focus, computer programming represents a relatively small share of use, at only 4.2% of messages.

  • Similarly, use cases related to companionship are niche; "Relationships and Personal Reflection" (1.9%) and "Games and Role Play" (0.4%) account for a very small fraction of conversations.

6. Education as a Core Use Case

A large share of ChatGPT activity revolves around learning and skill-building.

  • Tutoring or Teaching: 10.2 % of all messages

  • How-to Advice: 8.5 % of all messages

Together, these categories show that education is one of the biggest reasons people use ChatGPT - whether it’s formal tutoring, explaining concepts, or step-by-step guidance for practical task

Figure 9 - The important graphic in the study

You can read another good take on the study by Greg Isenberg here.

Watch The Next Wave Podcast

That’s it for this week’s Lore Brief.

See you next week!