From Billion Dollar Bets to Boardroom Shakeups

Ollie Carter

G'day and welcome back to The AI Intelligence Podcast, brought to you by Advancer - The AI Agency. It's Ollie here, joined by Llew Jury. Mate, this week in AI has been huge again. I thought we were due for a quiet patch, but nah, not a chance.

Llew Jury

Too right, Ollie. Hope everyone's got their caffeine sorted, 'cause there’s a heap to get through. If last week was a whirlwind, this one’s more like a cyclone for the AI industry. Let’s kick off with something that’s literally solar—you see that NASA and IBM Surya thing?

Ollie Carter

Yeah, Surya’s wild. Nine years of solar data they’ve managed to train into a foundational AI model. And we’re not just talking enterprise business tools, like, this is space science stuff—predicting solar storms, space weather, the works. I mean, when government and tech giants like IBM start collabing on AI for heliophysics, you know this isn’t about another chatbot.

Llew Jury

Exactly. Like, this kind of partnership—NASA and IBM pushing the edges of scientific discovery—reminds me of that conversation we had last episode about AI finding its way out of the corporate back office and into specialist fields. And look, the applications are massive: keeping satellites safe, understanding how the sun mucks with our comms here, even power grids.

Ollie Carter

Speaking of the back office, Cloudflare just rolled out numbers from their AI Week. Get this: their support agents now resolve 10 times as many customer tickets thanks to AI. I know companies love a good stat, but that's a drop-the-coffee-mug moment, isn't it?

Llew Jury

It’s next-level, mate. It backs up what we keep saying—AI isn’t just shiny PowerPoints developed by Copilot. When agents can handle 10 times the workload, that’s actual productivity, not just a line in a press release. But then, you see new research out of MIT? The flipside to all this hype?

Ollie Carter

Oh, the MIT study—yeah. People love to bang on about AI pilots sparking a revolution, but apparently 95% of those generative AI pilots at companies aren’t actually accelerating revenue. Only five percent are moving the needle. That’s an interesting stat if I ever heard one. I’ll admit, fell into that trap myself a few years ago trying to scale a chatbot—we had this great demo, everyone cheered at the board meeting, but when it went live? Barely a blip on the bottom line. Real-world implementation is just—different.

Llew Jury

Yeah, and that gap between "look at us, we're innovating" and "is anyone actually buying this?"—classic. The hype cycle chews up a lot of energy, but getting from pilot to profit is seriously tough.

Ollie Carter

I mean, it’s why stuff like Surya for NASA stands out. That’s years of effort, deep data. Not some quick hackathon output, right?

Llew Jury

Exactly. The industry’s starting to learn—real AI impact isn't just about rolling out pilots, but committing for the long run. Sort of a lesson in patience for everyone hyped about overnight transformation.

Llew Jury

And you can’t talk shifts without talking about the people behind the tech. Massive moves this week. OpenAI, for example—now they’re always in the headlines, but Julia Villagra stepping down as Chief People Officer after less than half a year? What's behind that you reckon?

Ollie Carter

Certainly an interesting one. She only took the role in March, promoted by Sam Altman himself, and now she’s out. That kind of turnover at the exec level usually means there’s much more rumbling beneath the surface. And you just know at OpenAI’s pace—scaling that fast—holding onto senior leaders must be like herding cats in a storm.

Llew Jury

Absolutely. When there’s this much money, press, pressure—turnover’s inevitable. And OpenAI’s not alone. Meta just made a huge move: hit pause on AI hiring so they can rethink their whole superintelligence plan. And at the same moment, they’re raiding Apple’s cupboard for talent, nabbing one of the senior Apple AI execs. It’s a proper talent war out there.

Ollie Carter

That’s the real arms race, isn’t it? Not just building the smartest AI, but getting the smartest people. Companies throwing everything into recruitment, pay bumps, crazy incentives. You remember your startup days, Llew? Platform lock-in nearly did my head in—if your people jump ship, the whole playbook goes with 'em, right?

Llew Jury

Yeah mate, I've been there, but it's how fast you bounce back. What’s wild now is how all these LLM platforms—OpenAI with GPT-5, Anthropic’s Claude Opus 4.1, Grok 4 from X, Gemini 2.5 Pro from Google—are all racing not just for performance but for staff who’ll keep them ahead. Context length has become the new specs war, but it’s the teams behind those models that really decide who stays top dog.

Ollie Carter

Absolutely. Gemini’s context length is nuts—up to a million tokens. That’s wild compared to the old days, right? For anyone not knee-deep in LLM jargon: longer context means more info in, more useful out, especially for tricky stuff. And companies have to figure out, "Do we need the flashiest model, or an open-source option we can actually control?" The whole landscape moves so fast, but open-source is speeding up too.

Llew Jury

Open-source is a big part of the competitive future, for sure. Gives smaller players a fighting chance, and lets devs jump between providers with things like AI Shell. Model-agnostic tools mean, hopefully, less lock-in headaches like we both remember.

Ollie Carter

The real challenge, though, is making these breakthroughs cost-effective and sustainable to run at scale. There’s this idea in the latest papers—test-time routing and efficiency tricks to get the best bang for buck from LLMs. Because businesses don’t just want the fanciest tech, they need to run it without busting their budgets.

Llew Jury

Spot on. It’s not just about big brains, it’s about big brains you can afford to run, hey? So the next winner might be the best balancing act, not just the flashiest benchmark scores.

Ollie Carter

Alright, let’s talk money—because if there’s one thing crazier than the tech, it’s the capital pouring in. OpenAI sniffing around a US$500 billion valuation. That’s five-zero-zero billion. Makes your eyes water just saying it.

Llew Jury

It’s ridiculous, isn’t it? And get this: Databricks just cracked the $100 billion mark, up a whopping 61% in eight months. These aren’t just numbers, this dwarfs entire industries. In the first half of 2025, AI startups raised a record $122 billion! You’ve got US deals absolutely dominating—$104 billion out of that total. It’s like the rest of tech has gone on pause just to pour money into AI start-ups.

Ollie Carter

Yeah, and there’s Field AI—robotics, AI-powered, now at a $2 billion valuation with backers like Bill Gates, Nvidia, Amazon… kind of makes you wonder if every VC on earth is just writing cheques with "AI" in the project description.

Llew Jury

If it’s not in your deck, you’re invisible to investors. But with all this capital, comes wild expectations—it's deliver or die. Which brings us to the spicier side—ethics and consent. Did you see the story about YouTube using AI to edit user videos without consent?

Ollie Carter

Yeah, that’s proper dystopian. Secret edits, no notice to the user. If it happened to my uploads, I wouldn't be happy. Raises the whole issue of who actually owns and controls your content in the AI era. And with tools that can quietly rewrite or manipulate anything, the whole idea of digital rights gets fuzzy—Orwell would have a field day with this.

Llew Jury

And it's only gonna get messier as AI gets more creative. On that note, did you catch the latest on AI “creativity”? There’s research dropping about how the hidden mechanics differ from human creativity, which—yeah, it’s a little academic, but it matters. The point is, the more AI can surprise us, the more we need to keep an eye on what that means for ownership and trust.

Ollie Carter

And speaking of surprises, a recent study used AI to chew through 7.9 million speeches and found that—wait for it—older adults are drivers of language change, not just the kids. Totally flips the script. Just like the industry, sometimes disruption comes from places you don’t expect.

Llew Jury

Yeah, and here’s a fun stat to wrap your brain around: 71% of organisations are using generative AI now. That’s not "maybe one day"—that’s here and now. The FDA’s cleared over 220 AI medical devices. And 150,000 autonomous vehicle rides, every week. The blurred lines aren’t just about who’s building AI, but literally how it’s shaping our day-to-day, in business and in culture.

Ollie Carter

So, looking at all these numbers, news and surprises, it’s clear this industry isn’t slowing down anytime soon. If anything, the gap between what’s possible and what’s responsible is only getting more interesting. That’s it for today’s episode, folks—next week there’ll be just as much drama, I reckon.

Llew Jury

Couldn’t agree more Ollie. And before we go, a shout and thanks to the great team and our sponsor Advancer. Advancer are the AI experts providing real world AI and Automation strategy, AI products and customised AI solutions for your business. Check them out at Advancer.com.au! Always a pleasure, Ollie.

Ollie Carter

Cheers, Llew, catch you then. Bye for now!

Llew Jury

Catch ya!