AI with Kyle Daily Update 066

Today in AI: OpenAI Dev Day

Author

Kyle Balmer
October 08, 2025

Catch Kyle’s latest video guide on Youtube.

Walking you through step by step:
How To Make $2,000 An Hour With AI Without Coding

The skinny on what's happening in AI - straight from the previous live session:

Highlights

💎 OpenAI's $63B AMD Deal: The Real Dev Day News

While everyone watched Dev Day, OpenAI quietly signed a deal with AMD for 6 gigawatts of GPUs, sending AMD's stock up 24% and adding $63 billion to their market cap in one day. OpenAI will also potentially acquire a large stake in AMD.

Kyle's take: This is more important than Dev Day but got buried. In the last couple of weeks OpenAI has reduced their reliance on both Microsoft (adding Oracle) and Nvidia (adding AMD) without spending a penny. It’s all based on future earnings and the money that the companies will make - together.

OpenAI have just rewritten the balance of the tech industry purely on future revenue promises. Everyone's now inextricably tied to OpenAI - they're becoming the linchpin of not just tech markets but the entire American economy. That's a lot of weight on one company. If this is a financial bubble it's going to be messy.

Source: Bloomberg

📱 ChatGPT Apps: The New App Store Gold Rush

OpenAI launched ChatGPT Apps with partners like Booking, Canva, Zillow, Spotify, DoorDash, and Khan Academy. SDK available now for developers, with an app store coming later this year. They demoed pulling Zillow data directly into Canva presentations through ChatGPT.

Canva app inside ChatGPT

Kyle's take: This has 2008 App Store vibes. Build something valuable, get it approved, and you have access to 800 million users. This could be the next goldrush for developers.

The GPT Store was a bust - nobody uses it, let’s be honest. And they didn’t introduce monetisation so what’s the incentive? But this is different because it's native integration, right in chat. And monetisation seems to be part of the plan (no details as of yet).

Start building NOW so when they open submissions later this year, you're ready. The monetisation model isn't clear yet (probably 30% like Apple? Perhaps?), but with shopping features coming, this could be massive. It opens up access to ChatGPT’s 800 million weekly users…right in ChatGPT.

UK/EU excluded initially, as usual. Comeon…

🤖 Agent Kit: Built in 6 Weeks with 80% AI Code

OpenAI's drag-and-drop agent builder looks exactly like n8n/Make/Zapier but locked to OpenAI models only. The kicker: they built it end-to-end in under 6 weeks with their Codex tool writing 80% of the code.

Kyle's take: Everyone's screaming "RIP n8n" but hold on. The massive limitation: you can only use OpenAI models. No Claude, no Gemini, nothing else. Just good ol’ ChatGPT. That kills flexibility for serious builders.

It's also weirdly positioned - too complex for beginners (no chat interface to help??), too simple for developers (drag and drop? Really?).

I think the big news here is the speed of deployment. If OpenAI can replicate n8n in 6 weeks with AI, what's anyone's moat anymore? I built my own course certification system yesterday in 3 hours instead of paying $150/month to some SaaS. This is coming for everyone.

Source: Steven Heidel

🎮 Sora Gets Reined In: No More Nazi SpongeBobs (BOO!)

After letting users go wild for virality (Sam Altman barbecuing Pikachu, anyone?), OpenAI's locking down Sora. That was quick! They're moving to officially licensed characters only, with revenue sharing for IP holders like Disney.

Kyle's take: We’ve seen this with lots of image and video model releases. Release unfettered for viral marketing (remember the studio Ghibli craze?) , then lock it down once everyone's hooked.

The licensing model sounds sorta like Fortnite skins - pay to use official characters. OpenAI have hinted at revenue sharing but no details yet. Will Disney et. al be interested? Or worth hanging onto their IP?

Source: The Verge

Member Question: "What are your thoughts on the cost funnel from AI users to AI companies to compute companies?"

Kyle's response: The revenue doesn't justify the valuations at all. There's not enough subscription or API revenue to sustain this - it all requires outside VC investment. It's not profitable yet.

The question is whether we can create enough economic value to justify current spend. This isn’t unusal in startups - Amazon made no profit for 12 years. But this time it’s a whole industry: AI. And arguably (with news of deals with Oracle and AMD) even larger than that. It’s potentially ALL of tech.

With AI, it's also about what we build on the application layer, not the base technology. Like steam engines - you don't necessarily make money on the engines, you make it on railways and factories built with them. That’s the next stage and the one that could make all of the investment economically viable.