No Image Available

AI at ‘Ludicrous Speed’: Anthropic Wolf Cries, OpenAI Price Wars & Amazon Borrowed Billions. ARD #95

Today’s theme: the AI industry’s perceived ‘Ludicrous Speed.’ Before anything else — watch the iconic 3+ minute ‘Ludicrous Speed’ clip from Spaceballs (1987). It’s the speed metaphor that’s run through the AI debate for four years — Elon literally put a Ludicrous Mode in the ‘plaid’ Tesla Model S. And this week Anthropic’s Dario Amodei reached for the opposite metaphor from his favorite fantasy epic: Lord of the Rings’ Treebeard, the Ent who moves at the speed of trees because hasty decisions last for centuries. From Anthropic’s exponential AI cries to Elon’s exponential space imaginations, with OpenAI in the middle trying to make a buck — three events today, each with my Take, and my Overall Take. On the eve of the SpaceX IPO.

All at ludicrous speed indeed.


(1) Anthropic’s Crescendo of AI Safety ‘Wolf’ Cries and Essays

MP TAKE: Dario has worn AI safety on his sleeve since the founding of his sibling AI company. He split from OpenAI four years ago on exactly this proposition, and the rest is history: Anthropic now races ahead of OpenAI on revenue, crossing $50 billion annualized. He’s the boy who cried wolf who truly believes the wolf came at every call — unlike Elon and Sam, who cry wolf more opportunistically. And here’s the thing: he isn’t wrong about the wolf out there. He’s describing the view from a watchtower the rest of us haven’t climbed. His few thousand researchers may have more AI compute per person than anyone on Earth, and they’re dog-fooding models that increasingly write more of themselves — companies spending hundreds of thousands+ of dollars of compute per researcher are startled by what these systems do when you open the throttle. They see things first.

But seeing first isn’t the same as seeing its impact across billions in the real world. More good is generally created from technology than bad. Case in point: Fire, despite its dangers. Also, the AI technology is already out in the world — the toothpaste is out of the tube — with hundreds of thousands+ of very smart people building on the same fundamentals, much of it open source, much of it in China. And when your latest ‘safe’ model won’t define mitochondria for fear the user is a budding bioterrorist, you’ve overshot the guardrail and handed the skeptics their argument. I believe Anthropic is exponentially under-estimating how technology actually diffuses through the real world — and how adaptable society is without their indulgent prodding. Adjusted, as always, for any underlying efforts at regulatory capture. To be clear: the safety work itself has real value. The debate is pace and policy — not whether to clip the models from delivering results for users.

Sources in Narrative order- The catalyst: Dario Amodei’s latest essay“Policy on the Exponential”, especially the opening paragraphs, plus his X post. Anthropic’s corporate essay on the same AI Exponential, Dario’s earlier optimist-counterpart “Machines of Loving Grace”, and the institute essay “When AI Builds Itself.” Then the backlash: TechCrunch on cybersecurity researchers unhappy with Fable’s guardrails, The Verge on Microsoft restricting Claude Fable 5 internally over the 30-day Mythos data retention, and The Verge again on “Claude Fable won’t answer basic biology questions.” For longtime readers, in narrative order: the self-appointed ‘Guardians of the AI Galaxy’ in ARD #54, Fable 5’s safety and pricing gates in today’s AI-RTZ #1114, Sam and Dario going opposite ways in AI-RTZ #1110, and ‘A Tale of Two AI Essays’ in AI-RTZ #508.


(2) OpenAI’s Counter-Pricing Moves for Market Share

MP TAKE : Sam Altman is playing the practical middle — and playing it well. His models are not far behind (GPT 5.6 is imminent, a real step past 5.5), his compute and financing are ahead of Anthropic’s today, and his IPO window is three to six months out, right on SpaceX’s heels. So when Anthropic pulls its best models out of even the $200-a-month subscription tier and goes a la carte, Sam goes the other way: token prices DOWN. That’s not charity — it’s the classic pre-IPO land-grab: spend margin on share now, sell the growth curve to public investors later. Especially when investors are more eager than wary. It also blunts the open-source flank from China and US players like Nvidia. The AI ‘glass half full’ strategy, against Dario’s ‘the glass is empty and about to explode.’ Both postures can be true at once. Only one of them takes market share vs competitors with more capital and distribution. What Amazon did well for decades, and what Google, Apple, Nvidia et al are fast emulating today.

Sources, in narrative order: The Information“OpenAI preps new AI model GPT 5.6”, signaling an IPO ‘within the next year.’ And the WSJ“OpenAI Considers Drastic Price Cuts, Anticipating War for Users with Anthropic.” For longtime readers: ‘It’s business AND personal’ in the pricing battles in AI-RTZ #1048.


(3) Amazon’s $17.5 Billion Loan — and Broader Implications

MP TAKE: Everyone’s back for seconds — and now thirds. Amazon already sits at the top of the AI capex leaderboard — roughly $200 billion this year, with Microsoft at ~$190 billion, Google at ~$170 billion, and Meta and Elon in the $120-billion-plus club — and now it’s filling up the gas tank with $17.5 billion of borrowed money, days after Google raised $80 billion in equity. Note WHO’S borrowing: a company with one of the strongest balance sheets on Earth taking cheap debt for scarce capacity is itself the tellJassy is borrowing while the lenders are eager. Because in this tech wave, unlike every prior one, the amount of compute gates everything — building the models, and getting them into the world where developers do things at 2x, 3x, 4x the pace otherwise possible. Every stakeholder in the AI Tech Wave is at the table. Andy Jassy is playing some of the most pragmatic hands at the AI poker table — tactically and strategically. Especially in recent close partnerships with both Anthropic and now shrewdly OpenAI.

Sources, in narrative order: BloombergAmazon inks a $17.5 billion loan led by Citigroup as the ‘AI borrowing frenzy’ grows. For longtime readers, in narrative order: Amazon AWS’s ‘no expense spared’ AI builds in AI-RTZ #999, yesterday’s ‘Everyone Ready for Seconds’ in ARD #94, the global rush to public-market funding in ARD #88, and Elon’s boundless AI ambitions in AI-RTZ #1066.


MP OVERALL TAKE

Step back, and the threads of the last hundred+ posts & episodes at AI-RTZ/ARD converge this week. The Guardians of the AI Galaxy are crying wolf from their watchtowers. The pragmatists are cutting prices to buy share before the IPO window. The incumbents are borrowing billions to fill the gas tank. To throttle faster, at their own ‘ludicrous speed’. The lights are flashing green, everyone’s back for bigger seconds — and SpaceX prices tonight, the largest IPO in history, with all of this happening at once on its eve.

MP Take: Three decades-plus of tech waves were a dress rehearsal for this moment. We are at far heightened levels of underlying technology change than any prior wave — I agree with the directionality of Anthropic’s pronouncements, while disagreeing with the glass-half-empty conclusions. Particularly because institutional and mainstream users are far more adaptable, smarter, and good than presumed. I ran internet research at Goldman in the nineties, and the lesson from the front row was unambiguous: the lighter regulatory touch on telecom and online commerce is what let the internet compound ahead of the rest of the world. The cost of the opposite error is on display right now: Europe’s preemptive AI rules mean 400-500 million people won’t even get Apple’s new Siri AI this fall — Tim Cook and Craig Federighi said so on stage. That’s what grabbing the brake too early costs. The other failure mode is Spaceballs’ — hit Ludicrous Speed, overshoot the mark, and end up motion-sick in deep space. Treebeard speed, general human speed, Ludicrous Speed: the instrument that needs mastering isn’t the brake or the afterburner. It’s the throttle. This year and next.


Gadget AI — Apple’s Siri AI Seems ‘Good Enough’ in Early Beta

Sources, in narrative order: The Verge’s detailed early beta review — “I tried Siri AI, and so far it actually works” — trustworthy, factual, no hallucinations, working bottom-up in the apps. For longtime readers: Apple can win AI on the iPhone in AI-RTZ #1113.

MP Take: The Apple iPhone may be the AI Device we’ve all been waiting for. Just like ‘Intel Inside’ made the PC mainstream, it’s ‘Google and Nvidia Inside’ that may make Siri AI the mainstream AI — in billions of pockets, no new gadget required, just updated software. A very different picture than three to six months ago, when the world was complaining Siri was late (it was) and not up to the job (it wasn’t). The early verdict is ‘good enough’ — and in consumer tech, ‘good enough’ at iPhone scale beats brilliant in a demo, every time. Ask Netscape: 80% browser share to under 20% in less than two years once Microsoft’s good-enough browser shipped with the operating system. Customer reach wins every time. That’s the ludicrous speed worth watching — distribution, not demos. The AI device race may end the way it started: with the phone already in your hand. Just a little more memory.


Questions

Q1 — What’s the most practical aspect of AI MP is most excited about over the next 18 months?

‘Good enough’ AI — from Siri AI and Apple Intelligence, and from Google’s Gemini and AI Mode in Search. Mostly free, capable with mainstream requests, and easy training wheels to AI agents via shortcuts, extensions and app intents. Good-enough with distribution gets underestimated every tech wave — and it tends to win.

Q2 — What’s MP’s most practical concern on AI pricing?

That a la carte pricing delays general recognition of the value of AI and AI agents — thus postponing the net good AI possibilities. The faster mainstream users feel the value, the faster the flywheel turns.


Full Source Reading —

For the broader context, see the canonical sources for ARD 95 — in today’s narrative order:

Preamble — the speed metaphors

Event 1 — Anthropic’s Safety ‘Wolf’ Cries

Event 2 — OpenAI’s Counter-Pricing

Event 3 — Amazon’s $17.5B Loan

Gadget AI — Siri AI in Beta


Shorts Clips from today

Clip 1 — Siri AI: Good Enough, In Billions of Pockets

Watch on YouTube Shorts

The Verge’s early review of Apple’s Siri AI beta says it all: “I tried Siri AI, and so far it actually works.” Trustworthy, factual, doesn’t hallucinate — and it works bottom-up in the apps. The Apple iPhone may be the AI device we’ve all been waiting for.

MP Take: Just like ‘Intel Inside’ made the PC mainstream, it’s ‘Google and Nvidia Inside’ that may make Siri AI the mainstream AI — in billions of pockets, no new gadget required, just updated software. In consumer tech, ‘good enough’ at iPhone scale beats brilliant in a demo, every time.

Clip 2 — Dario’s AI Wolf Cries from the Watchtower

Watch on YouTube Shorts

Anthropic’s Dario Amodei has worn AI safety on his sleeve from the founding — it IS the founding proposition of the company. The ‘boy who cried wolf’ who may truly believe the wolf came at every call, with researchers living in an exponential future on more internal AI compute than possibly any company worldwide.

MP Take: Dario isn’t lying about the wolf — he’s describing the view from a watchtower the rest of us haven’t climbed. But Anthropic is exponentially under-estimating how technology diffuses in the real world, and society’s adaptability without their prodding. The safety work has real value — the debate is pace and policy, not whether.

Clip 3 — Every Tech Wave Was a Dress Rehearsal for AI

Watch on YouTube Shorts

Three decades of tech waves — the PC, the internet, mobile — were a dress rehearsal for this. The 1990s internet showed what a lighter regulatory touch makes possible: bottom-up innovation compounding ahead of the rest of the world.

MP Take: I disagree with the ‘glass half empty’ proclamations, but institutional and mainstream users are far more adaptable than presumed. The instrument that needs mastering isn’t the brake or the afterburner. It’s the throttle.

Clip 4 — The AI Speed Gap: Researchers vs the Rest of Us

Watch on YouTube Shorts

There’s a growing gap between the AI improvement the rest of us see and what frontier researchers see — companies spending hundreds of thousands of dollars of compute per researcher are startled by what these systems do when you open the throttle. That gap is the heart of Dario’s ‘country of geniuses in a data center.’

MP Take: Maybe so — but the toothpaste is out of the tube. The technology is in the real world, with hundreds of thousands of very smart people building on it, much of it open source. Guardrails yes — but the question is pace and policy at what speed, not whether.


AI Ramblings Daily on AI-RTZ is here to think through AI and reset. Together.

Today’s AI-RTZ #1114 — Anthropic recasts Mythos as Fable 5, with safety & pricing gates — the ‘safer and pricier’ gating of the latest fabled models, ahead of the mega-AI IPO. A lot more detail there on what I discussed today.

Tomorrow — ARD 96 on AI-RTZ #1115.

Thanks for tuning in, AI Curious Folk. Stay tuned.

(NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here.)

Links

Preamble — speed metaphors

  • Spaceballs ‘Ludicrous Speed’ clip (3:30):

Event 1 — Anthropic’s Safety ‘Wolf’ Cries (narrative order)

AI: Reset to Zero
Anthropic & OpenAI, the self-appointed ‘Guardians of the AI Galaxy. ARD #54
Read more

  • AI-RTZ #1114 — Anthropic recasts Mythos as Fable 5:

AI: Reset to Zero
AI: Anthropic recasts Mythos as Fable 5, with safety & pricing gates. AI-RTZ #1114
Anthropic has finally released ‘Fable 5’, its much heralded and renamed Mythos model in a ‘safer’, and pricier, ‘a la carte’, version than its previous Claude Opus 4.8…
Read more

  • AI-RTZ #1110 — Sam and Dario go opposite ways:

AI: Reset to Zero
AI: Sam & Dario counter-stir the ‘AI Fears’ Pot. AI-RTZ #1101
I’ve long discussed the current mainstream fears around AI, its decade old origins that led to OpenAI. Along with its sibling Anthropic AI. As a result, how both are rooted in the doomer vs accelerationist AI crowds…
Read more

  • AI-RTZ #508 — A Tale of Two AI Essays:

AI: Reset to Zero
AI: A Tale of Two AI Essays. RTZ #508
The Bigger Picture, October 13, 2024…
Read more

Event 2 — OpenAI Counter-Pricing (narrative order)

AI: Reset to Zero
AI: It’s Business and Personal in OpenAI/Anthropic pricing battles. RTZ #1048
Most of the time it’s both personal and business. Especially as one goes through the details in detail. Between sibling rivals, Anthropic and OpenAI I mean…
Read more

Event 3 — Amazon $17.5B Loan (narrative order)

AI: Reset to Zero
AI: Amazon AWS’s ‘No Expense Spared’ AI Builds. RTZ #999
I’ve written a lot for a while now, on how Amazon is poised to be a strong big tech beneficiary of the early AI Tech Wave. In particular with its crown jewel Amazon Web Services (aka AWS) Cloud business…
Read more

  • ARD #94 — Everyone Ready for Seconds:

AI: Reset to Zero
Everyone Ready for Seconds: AI’s Wall of Worry Recedes. ARD #94
Read more

  • ARD #88 — Global Rush to Public Market Funding:

AI: Reset to Zero
Global Rush to Public Market Equity Funding. ARD #88
Read more

  • AI-RTZ #1066 — SpaceX/xAI’s Boundless AI Ambitions:

AI: Reset to Zero
AI: Spacex/xAI IPO filing outlines Elon’s boundless AI Ambitions. AI-RTZ #1066
So many AI balloons being floated in Elon Musk’s SpaceX/xAI confidential IPO S-1 filing. Especially driven by xAI/Grok’s AI priorities over SpaceX/Starlink. Let’s unpack…
Read more

  • AI Tech Wave evergreen (inline in Take 3):

AI: Reset to Zero
AI: Building Value over Time
Over the last thirty plus years, each major technology wave, like the PC and then the Internet, evolved as a series of technologies in a tech value stack that came to define the full ecosystem with huge collective value over time. They then went on to create the winning companies in each stack worth billions, and some now in the trillions…
Read more

Gadget AI + Questions (narrative order)

AI: Reset to Zero
AI: Apple Intelligence & Siri AI hum with ‘Google & Nvidia’ inside. AI-RTZ #1113
Paraphrasing an iconic Intel campaign, it’s Apple’s AI strategy, with ‘Google & Nvidia Inside’. Let me unpack…
Read more

Today’s companion post + episode + clips

  • AI-RTZ #1114 — Anthropic recasts Mythos as Fable 5 (today’s companion):

AI: Reset to Zero
AI: Anthropic recasts Mythos as Fable 5, with safety & pricing gates. AI-RTZ #1114
Anthropic has finally released ‘Fable 5’, its much heralded and renamed Mythos model in a ‘safer’, and pricier, ‘a la carte’, version than its previous Claude Opus 4.8…
Read more

  • ARD 95 — Main on YouTube:

  • Short 1 — Siri AI: Good Enough, In Billions of Pockets:

  • Short 2 — Dario’s AI Wolf Cries from the Watchtower:

  • Short 3 — Every Tech Wave Was a Dress Rehearsal for AI:

  • Short 4 — The AI Speed Gap: Researchers vs the Rest of Us:


Subscribe to AI: Reset to Zero for daily AI Ramblings + Sunday Bigger Picture posts.





Want the latest?

Sign up for Michael Parekh's Newsletter below:


Subscribe Here