Latest / AI Fire Daily / 🎙️ EP 260: The $10B AI Takeover & Hand-Controlling Wearables
Transcript
- 0:00For years, the browser was this perfectly safe
- 0:03sandbox. We kept our AI models neatly contained,
- 0:07you know, behind digital screens. Yeah, totally
- 0:10contained. But right now, the tech is aggressively
- 0:13stepping out. It's literally taking physical
- 0:16control of the real world. We're watching it
- 0:19run the back end of major banks. We're even seeing
- 0:22it... um control human hand muscles directly
- 0:26which is just crazy i mean we are talking about
- 0:28raw electrical signals bypassing our brains entirely
- 0:31so welcome to today's deep dive i'm really glad
- 0:35you're here with us yeah thanks for having me
- 0:37today we're looking at the landscape of may 2026.
- 0:41we are unpacking a massive shift happening right
- 0:43now ai labs are just Well, they're done waiting
- 0:46for users to sign up. They're completely done.
- 0:49Right. They're forcibly embedding themselves
- 0:50into the global economy. They're entirely rebuilding
- 0:53their infrastructure just to handle real time
- 0:57human interaction and, you know, causing real
- 0:59world hardware shortages in the process. It's
- 1:02an incredibly wild time to be watching this space.
- 1:04The whole landscape is shifting right under our
- 1:06feet. We're seeing this rapid transition from
- 1:09like digital toys to. physical reality. And it's
- 1:13happening much faster than anyone modeled. Exactly.
- 1:16So let's start by just following the money. Always
- 1:18a good idea to understand where AI is going.
- 1:20We first have to look at the enterprise side.
- 1:22The biggest labs in the world are forcing their
- 1:24way in. They're actively drilling into the corporate
- 1:27bedrock of our economy because they have these
- 1:30massive IPO goals they need to hit. And to do
- 1:34that, they basically need to be indispensable.
- 1:35Right. And. Subscriptions alone just aren't enough
- 1:39anymore. OpenAI and Tropic realized a very hard
- 1:44financial truth recently. Yeah. You simply can't
- 1:46wait for individual people to buy a plus plan.
- 1:49Consumer subscriptions are just historically
- 1:51fickle. Users churn all the time. They do. If
- 1:55you want to hit... multi -billion dollar IPO
- 1:57numbers, you need guaranteed revenue. You need
- 2:00deep institutional contracts that span years.
- 2:03You essentially have to physically go into their
- 2:05offices. You have to custom build the intelligence
- 2:07for them. OpenAI just made a structural move
- 2:10here. They quietly created something called the
- 2:13deployment company. Oh, yeah. It's already valued
- 2:15at. $10 billion. Which is staggering. Right.
- 2:20OpenAI owns the vast majority of the equity,
- 2:22but they've got serious heavy hitters backing
- 2:25them. TPG, Brookfield, SoftBank, they're all
- 2:28getting involved. That is undeniably serious
- 2:30institutional capital entering the chat. And
- 2:33their deployment strategy is, frankly, highly
- 2:37aggressive. They're targeting over 2 ,000 major
- 2:40portfolio companies. The ultimate goal is integrating
- 2:43GPT 5 .4 and Codex directly into their systems.
- 2:47They want to fundamentally rewire how these companies
- 2:50operate. It's not a tool anymore. It's a completely
- 2:52new operating system. A totally new operating
- 2:54system. Yeah. And Anthropic certainly isn't sitting
- 2:56on the sidelines here. They formed their own
- 2:59private equity coalition to compete. Right. Blackstone,
- 3:02Goldman Sachs, Hellman and Friedman. Anthropic
- 3:04is aggressively going after the midsize market.
- 3:06But, you know, their actual integration method
- 3:08is fascinating to me. It really is. They aren't
- 3:11just sending a download link. They're literally
- 3:14embedding their own flesh and blood engineers
- 3:16into these companies. Right. And that is a crucial
- 3:18distinction. They aren't just selling an API
- 3:20key and hoping for the best. No. They're sending
- 3:24highly paid human engineers physically into the
- 3:27building. These engineers are actively rewriting
- 3:30entrenched legacy workflows. They're using Claude
- 3:34to rebuild operations from scratch. So they're
- 3:37basically hiring a totally new kind of worker,
- 3:39not just developers who sit in code. They need
- 3:41engineers who can persuasively talk to a CEO.
- 3:45Yeah, translators, basically. Exactly. The biggest
- 3:47example of this is Anthropic's partnership with
- 3:50FIS. FIS is a staggeringly massive deal. For
- 3:54context, it's a software running a huge chunk
- 3:57of global banks. We're talking about the core
- 3:59financial plumbing of the world. Anthropic is
- 4:02bringing Claude directly into that sensitive
- 4:04ecosystem. By late 2026, it's going to be widely
- 4:08available across the sector. Institutions like
- 4:10BMO and Amalgamated Bank will be running on it.
- 4:13It's kind of like stacking Lego blocks of data
- 4:15directly into the foundation of Wall Street.
- 4:17That's a great way to put it. But these blocks
- 4:19are constantly thinking. They're actively adapting.
- 4:22If one block decides to change its shape, the
- 4:24whole tower shakes. Beat. It's honestly a little
- 4:29terrifying. Oh, absolutely. I mean, I still wrestle
- 4:31with prompt drift myself. We all do. Getting
- 4:34a model to just stay on track is genuinely hard.
- 4:37Yet they're trusting AI to run global banks seamlessly.
- 4:41Well, that underlying risk is exactly why they
- 4:43embed human engineers. They absolutely cannot
- 4:46afford any prompt drift at a major bank. Right.
- 4:50Joint ventures secure that guaranteed revenue
- 4:52stream. When you're deeply integrated into a
- 4:55bank's ledger, you're incredibly sticky. You
- 4:57don't just get canceled like a Spotify subscription.
- 4:59No, you become a permanent utility. Exactly.
- 5:03Sierra is another perfect example of this enterprise
- 5:05demand. Yeah. Sierra's wild. They just raised
- 5:08$950 million. Yeah. That puts their valuation
- 5:11at $15 billion. They already serve 40 % of the
- 5:14entire Fortune 50. Their annual recurring revenue
- 5:17jumped from $100 million to $150 million. Almost
- 5:20overnight. It's crazy fast. It clearly proves
- 5:23the corporate market is starving for this infrastructure.
- 5:26It effectively shows that enterprise AI is not
- 5:29a bubble. It's a structural rewiring of how corporate
- 5:33infrastructure operates. The raw demand for automated
- 5:36workflows is just staggering. Companies are panicking
- 5:40that they'd be left behind. Okay, let me pause
- 5:42and get this straight. If they are embedding
- 5:44AI into the core of global banking, how does
- 5:47the infrastructure keep up without crashing?
- 5:50Well, they realized they couldn't use the old
- 5:52Internet backbone. They had to completely rip
- 5:54out the old web architecture. They built a totally
- 5:56new split brain system for real time scale. So
- 6:00they basically built a brand new Internet just
- 6:02to handle the load. Exactly. And that naturally
- 6:05brings us to the voice infrastructure miracle.
- 6:07You simply can't rewrite global bank operations
- 6:10without flawless systems. You need split -second
- 6:13reliability at a massive scale. OpenAI recently
- 6:15released an engineering deep dive detailing this
- 6:18exact flex. And the raw numbers are absolutely
- 6:20staggering. As of May 2026, OpenAI officially
- 6:24handles 900 million weekly active users. Think
- 6:27about the physical volume of data moving there.
- 6:30To make a digital conversation feel human, it
- 6:33has to be fast. The AI constantly has to handle
- 6:36unpredictable human interruptions. It has to
- 6:39manage conversational turn -taking seamlessly.
- 6:41And traditional web setups just couldn't handle
- 6:44it. that kind of dynamic load. Yeah. Standard
- 6:45HTTP requests naturally have way too much inherent
- 6:48lag. It breaks the illusion of a conversation
- 6:51instantly. Yeah, it does. So OpenAI just ditched
- 6:54the old setups entirely. They moved to a custom
- 6:57WebRTC on Kubernetes stack. Let's define that
- 7:00jargon quickly. That's just tech that keeps live
- 7:03audio streams stable without crashing. Spot on.
- 7:06For you listening, think of the old internet
- 7:08like sending letters. There's always a noticeable
- 7:10delay waiting for a response. WebRTC is like
- 7:14keeping an unclosable phone line directly open.
- 7:17That's a perfect visual. And to actually make
- 7:19it work, they created a brilliant split -brain
- 7:22infrastructure. They didn't just shove everything
- 7:24onto one overheating server. They divided the
- 7:27computational labor to mass. maximize speed.
- 7:29Right. They built a very lightweight component
- 7:31they call a relay. The relay just handles the
- 7:34fast moving data packets. Its only job is to
- 7:36move raw audio quickly. Then. They built a much
- 7:41heavier stateful transceiver. The transceiver
- 7:43does all the actual heavy cognitive lifting.
- 7:45It handles the complex AI thinking and all the
- 7:48deep encryption. And they aggressively pushed
- 7:50this entire architecture out to the edge. They
- 7:53systematically deployed global relays at the
- 7:55Internet's physical edge. Which is wildly expensive.
- 7:58Incredibly expensive. It effectively means your
- 8:01voice hits a server almost instantly. The millisecond
- 8:04a sound leaves your lips, it... It aggressively
- 8:06cuts down the jitter and the conversational lag.
- 8:09It happens before the data even reaches the main
- 8:12model. And, you know, the core models themselves
- 8:14are entirely different now. We're talking about
- 8:17GPT 5 .5 and GPT real -time 1 .5. They are entirely
- 8:21audio native from the ground up. This is a profoundly
- 8:25deep shift in computer science. They aren't translating
- 8:28your natural speech into text anymore. Instead
- 8:30of reading a transcript, they're processing the
- 8:32pure sound waves. They can actually hear the
- 8:34hesitation or excitement in your voice. They
- 8:36process pure sound and speak with raw emotion
- 8:39directly. The median latency is currently sitting
- 8:42under 500 milliseconds. That is literally faster
- 8:45than human reaction time. Whoa. Imagine scaling
- 8:49to a billion queries. Two sec silence. It's genuinely
- 8:55hard to even wrap your head around the physics
- 8:57of that. Managing millions of concurrent stateful
- 9:00audio sessions is computationally brutal. Stateful
- 9:04means the server constantly remembers the entire
- 9:06conversation. It holds the context open. Exactly.
- 9:09It holds the context completely open while you
- 9:11talk. Doing that seamlessly for 900 million people
- 9:14without dropping context is brilliant engineering.
- 9:16But hold on. Think about the real world impact
- 9:18here. With 900 million people talking to audio
- 9:21native AI in real time, isn't that putting insane
- 9:24stress on actual physical hardware? Oh, it absolutely
- 9:27is. It's heavily straining the global power grid.
- 9:30It's gotten to the point that developers are
- 9:32panic buying consumer hardware to keep up. Wow.
- 9:35The cloud alone simply can't handle everything
- 9:37locally anymore. Right. The digital boom is creating
- 9:40a massive physical supply chain bottleneck. Exactly.
- 9:44And that seamlessly leads us to the profound
- 9:47fiction we're seeing. We are actively moving
- 9:49from digital infrastructure into physical reality.
- 9:52The scale of pure audio models. hits the real
- 9:56world hard. So if OpenAI is processing 900 million
- 9:59live audio streams, they're shifting a massive
- 10:02compute burden onto the grid. And that macroeconomic
- 10:06cloud strain is why we're seeing a microeconomic
- 10:09panic at the consumer level. Specifically, we
- 10:12are seeing this happen with Apple hardware. Yeah,
- 10:14Apple didn't foresee this. They didn't completely
- 10:16foresee this sudden crunch. Right now, Mac Mini
- 10:18and Mac Studio stock is running critically low.
- 10:21Software developers are aggressively hoarding
- 10:24these specific desks. cop machines because they
- 10:26desperately need them to run their local AI agents.
- 10:29Prices on the secondary market are already rising
- 10:31significantly. People intensely want to run AI
- 10:34locally for strict data privacy. They want to
- 10:37avoid that frustrating cloud latency entirely.
- 10:39So they're literally buying up every single Mac
- 10:42they can find. Industry experts think this severe
- 10:45shortage could easily hit iPhones next. If the
- 10:48iPhone supply chain actually gets hit, that is
- 10:51massive. It changes consumer tech availability
- 10:54overnight. It really does. And it's definitely
- 10:56not just physical supply chain friction we're
- 10:58dealing with. We're actively seeing major regulatory
- 11:01friction popping up, too. The White House is
- 11:03currently weighing some very serious new AI oversight
- 11:07rules. Yeah, they urgently want early access
- 11:09to new foundational models. The stated rationale
- 11:13from the administration is actually pretty straightforward.
- 11:15They want to manage the societal risks if things
- 11:18go completely wrong. It isn't really about broadly
- 11:21blocking AI development. They just want a thorough
- 11:24look under the hood before public deployment.
- 11:26Wait, hold on. If I'm understanding this dynamic
- 11:28right, the White House isn't trying to shut these
- 11:31local agents down. They just want a backdoor
- 11:32view into them because they're nervous. They're
- 11:34realizing an uncensored AI running freely on
- 11:37a local Mac studio is a wild card. Is that basically
- 11:41what's driving this sudden oversight push? Yeah,
- 11:44because the sheer lack of visibility is deeply
- 11:46concerning to regulators. If we connect the dots,
- 11:49the entire trend makes sense. The intense consumer
- 11:52desire for privacy and edge computing is driving
- 11:55this shift. Right. Everyday people want autonomous
- 11:58agents running locally in their own homes. That
- 12:01desire directly drives the massive Apple hardware
- 12:04shortage. And that decentralized deployment immediately
- 12:08prompts governments to seek visibility. They
- 12:10desperately want to know what is actually running
- 12:12on those machines. But let's bring this down
- 12:14to the listener for a second. If governments
- 12:15and hardware can barely keep up. How is this
- 12:18rapid integration actually showing up for everyday
- 12:20users right now? It's actively bleeding into
- 12:22highly personal physical applications at an unprecedented
- 12:26pace. We're seeing it fundamentally shift physical
- 12:29wearables and daily digital tools. So the tech
- 12:32is already fundamentally altering our daily physical
- 12:35and digital routines. Completely. We're rapidly
- 12:38moving from macroeconomic shortages down to the
- 12:40micro level. This is fundamentally about what
- 12:42you can actually use today. The sheer speed of
- 12:46practical applications application development
- 12:47is staggering right now. Let's actually look
- 12:50at the wildest edge case first, because this
- 12:52one honestly blew my mind when I read about it.
- 12:54There's a brand new AI wearable device out there
- 12:57right now. Well, this is crazy. It literally
- 13:00controls your human hands using direct electrical
- 13:02signals. It actively lets you perform physical
- 13:05skills you never learned. It's fascinating. It
- 13:08sends calibrated impulses straight to your hand
- 13:11muscles. It's almost literally like downloading
- 13:14physical abilities directly into your body. Yeah.
- 13:18It bypasses your brain's slow motor learning
- 13:20process entirely. It feels very sci -fi. It feels
- 13:23exactly like that scene in The Matrix. Totally.
- 13:25You just plug a cable in and suddenly you magically
- 13:27know Kung Fu. Or, you know, maybe you instantly
- 13:29know how to play the piano. You strap it on.
- 13:31Yeah. Your fingers are doing complex tasks. It's
- 13:34fascinating, but it's also profoundly weird to
- 13:38think about. It definitely bridges a crazy gap
- 13:40between biology and machine. But we also have
- 13:44very practical everyday tools bridging a similar
- 13:47gap. Look at what XAI just launched into the
- 13:50market this week. They released a wildly powerful
- 13:54new API specifically for voice cloning. It's
- 13:58wild. It lets you create. hyper -realistic custom
- 14:01voices. You can use them for podcasts, autonomous
- 14:04agents, synthetic videos. You can freely pick
- 14:06from over 80 distinct voices. They cover 28 different
- 14:10global languages with perfect accents. And the
- 14:13actual pricing model is what's truly disruptive.
- 14:15It starts at literally just $3 an hour. It completely
- 14:18commoditizes high -end voice production. Yeah.
- 14:20Anyone with a laptop can instantly spin up a
- 14:23multilingual ad campaign now. You don't need
- 14:25a massive recording studio or professional actors
- 14:27anymore. Right. Speaking of the advertising world,
- 14:30Meta just made a huge integration play. Meta
- 14:32now actively lets you connect AI straight into
- 14:35your ad account. You can directly plug in ChatGP
- 14:38to your cloud to run campaigns. The AI natively
- 14:41talks to your potential customers directly in
- 14:43the chat. The rapid adoption rate there is absolutely
- 14:45staggering. Weekly automated conversations jumped
- 14:48from 1 million to 10 million almost overnight.
- 14:51Wow. Small businesses are entirely letting the
- 14:53AI handle their sales funnels. It's negotiating,
- 14:56answering questions and closing deals in real
- 14:59time. We're also seeing an absolute flood of
- 15:02rapid fire. Daily tool updates, just highly practical
- 15:06tools you might use every single day. There's
- 15:08a new one called Avatar getting a lot of traction.
- 15:11It perfectly removes complex image backgrounds
- 15:14in one single click. It automatically balances
- 15:16colors flawlessly. It even intelligently restores
- 15:20missing parts of an old image. Then you have
- 15:22highly creative niche things like codex pets.
- 15:25I love them. Right? They're totally optional,
- 15:28small, animated companions specifically for Codex.
- 15:31They just sit quietly on your screen and show
- 15:33your thread status. They physically reflect whether
- 15:36Codex is actively running or just waiting. It
- 15:38has an interesting bit of personality to the
- 15:40coding process. There's also Droppy, which I
- 15:44think is a massive shift in retail power. It
- 15:46isn't just a basic price tracker. It's an autonomous
- 15:48agent fighting massive retail pricing algorithms.
- 15:51It constantly tracks Amazon, eBay, and AliExpress
- 15:55silently in the background. You just get a notification
- 15:57the exact millisecond a price drops. It shifts
- 16:00the power completely back to the consumer. And
- 16:02for independent website owners, there's a huge
- 16:04shift with sleek analytics. It's a completely...
- 16:08Privacy -first encrypted alternative to Google
- 16:11Analytics. It aggressively offers real -time
- 16:13data, entirely cookie -less tracking, and fast
- 16:16dashboards. All these diverse tools share one
- 16:19major common thread. They're taking highly complex
- 16:23AI architecture and making it utterly simple
- 16:25to use. Let's pause on that idea for a second
- 16:27because the societal implications are massive.
- 16:29With AI literally moving our hands and cloning
- 16:31our voices, where does the human element fit
- 16:33in? I think the human element fundamentally shifts
- 16:36up the cognitive chain entirely. As AI flawlessly
- 16:39handles the raw execution of tasks, we elevate.
- 16:42Our primary role permanently becomes one of deep
- 16:46curation. We handle the broad strategy and the
- 16:49complex intention behind the action. We decide
- 16:52what needs doing and the AI natively does it.
- 16:54Basically, we stop doing the heavy lifting and
- 16:57become the directors. Exactly. We confidently
- 16:59call the strategic shots. We architect the vision
- 17:01and the machine executes the labor. Let's pull
- 17:04all of this incredible information together.
- 17:06We've covered a truly massive amount of ground
- 17:09today. Mid -roll sponsor Red Placeholder. We
- 17:12started by looking at AI labs infiltrating major
- 17:15banks. They launched a massive $10 billion integration
- 17:19strategy to do it. They're actively securing
- 17:21guaranteed revenue from the global corporate
- 17:23economy. Then we look deep into the underlying
- 17:25infrastructure, making it possible. They absolutely
- 17:28had to rebuild internet audio architecture from
- 17:30scratch. Yeah. They're serving 900 million users
- 17:33seamlessly with a split brain system. They completely
- 17:36achieved sub -500 millisecond latency for pure
- 17:40emotional audio processing. And that massive
- 17:43digital explosion inevitably caused serious physical
- 17:47friction. We saw severe Apple hardware shortages
- 17:51hitting the secondary market hard. Software developers
- 17:54are aggressively hoarding Macs to run local agents.
- 17:57And the White House is pushing hard for new regulatory
- 18:00oversight. And it all eventually culminates in
- 18:03the most personal edge cases imaginable. Wearables
- 18:06that literally control your human muscles with
- 18:09electrical signals. XAI flawlessly cloning your
- 18:12exact voice for just $3 an hour. The great integration
- 18:16is officially no longer just a pending software
- 18:19update. It is a tangible, rapidly accelerating.
- 18:23physical reality happening right now. It really
- 18:25truly is. Thank you so much for joining us on
- 18:26this deep dive today. I highly encourage you
- 18:29to actively check out some of the specific tools
- 18:31we mentioned today. See exactly how they might
- 18:33seamlessly fit into your own daily workflow.
- 18:35And please be sure to subscribe for our next
- 18:38deep dive into this rapidly changing world. It's
- 18:40always an absolute pleasure to critically unpack
- 18:42this landscape with you. There's literally always
- 18:44something entirely new and profound to learn.
- 18:46I want to leave you with one final lingering
- 18:48thought today. If AI can perfectly replicate
- 18:51our voices with XAI and literally control our
- 18:54physical movements with new wearables, how long
- 18:57until we can't tell the difference between a
- 18:59skill we genuinely learned and a skill we simply
- 19:01rented for the afternoon? Beat. Let that sink
- 19:04in.