Dario Amodei shipped Opus 5 cheaper than Fable 5, Jensen Huang bet $500 billion on Korea, and Trump's AI czar resigned before unpacking his desk. Is the AI race even the same race anymore?
Hi,
My buddy Amos just open-sourced something worth your time.
It's called GTM Skills. First marketplace of its kind. 200+ human-crafted skills your AI agents can actually run. Free.
The premise is simple. Your agent is only as good as what you feed it. Most people feed it nothing, then act surprised when it writes like a robot with a quota.
So they went and got the actual operators.
Manny Medina built Outreach to $4.4B. His AI playbook is a skill now.
Sangram Vajre invented ABM and took two companies past $100M. His category creation framework is a skill.
Kevin Dorsey's sales methodology, the one 148K people follow, is a skill.
Alex Vacca built ColdIQ to $7M+ ARR as the most AI-native GTM agency out there. His entire library is installable.
Prospecting, signals, outreach, SEO, ads, ABM, deal acceleration, cold calling, pricing, CRM cleanup. Pick a task, there's a skill for it.
The stated goal is killing AI slop. The generated comments. The spam sequences. The blog posts that say nothing at all.
I'm in on that. You can find the skill I contributed here.
Works with Claude, ChatGPT, Swan AI, or whatever agent you're running.
It’s free. Like free free.
Go take a look. Or contribute one.
Let’s go.
This week from the AI for Founders podcast on Spotify :
Fourteen terminals, zero leverage — LINK
Ilan Man of Paradox Machines spent four months out of a two-decade data career building the company mid-market and PE-backed businesses actually need, and insists real leverage means sustainable systems, not always-on agent swarms.
The GDPR rule nobody remembered to delete — LINK
Matt Hassett of Loftie found a years-old email rule silently blocking half his new customers, so he built Deliberate, a ten-agent AI team that runs his ops for $7,500 a month.
Humans have a fifth drive, and most AI buries it — LINK
Jason Katz of Kindling Solutions says the morning brief is the best entry point into AI and also the ceiling for any company that hasn't fixed its operational leadership first.
He sold his last company for $217 million, then went looking for the next fight — LINK
Tarun Raisoni of Gruve argues today's AI pricing is investor-subsidized and temporary, and that the real fight ahead is over who owns the intelligence hiding inside your proprietary data.
The world’s best podcast player. Snipd
Investor Corner
Are you raising? Reply here, and we’ll feature your deck.
An autonomous AI job search agent for senior tech professionals. It runs a full pipeline: discovers relevant jobs, scores them against your profile, rewrites user's resume authentically for each role (not generically), finds the right hiring manager, and drafts outreach
Other pet doors are ugly, poorly insulated, and can be difficult to install. Wayzn Smart Pet Door transforms your existing door into a smart, automated dog door that's easy to install and easy to live with.
DiningOut unites restaurants, food lovers, industry suppliers, and brands through a self-reinforcing ecosystem of live events, targeted media, and proprietary technology.
Rare, Pristine, Protected drinking water.
The only platform that can authenticate human-made content. Using patented algorithms that analyze how content is made, not just the output.
Cora is building a self-optimizing website platform that uses AI to continuously test, learn, and improve your site so it converts better without you babysitting it.
AI-powered training simulations to help healthcare learners practice complex interpersonal and communication skills. We focus on realism, repeatability, and scalability—making soft-skills training more accessible and measurable.
AI + Human Stylists → LEARN MORE → Save $30 with RYAN30
Tip of the spear:
Anthropic settles its $1.5 billion copyright case — LINK
A federal judge gave final approval to Anthropic's landmark settlement with authors and publishers, paying roughly $3,000 per work across 500,000 titles. The ruling that AI training counts as fair use stands, but it's not binding precedent, so this fight is far from over industry-wide.
Anthropic's Opus 5 costs way less than Fable 5 and still beats it on benchmarks — LINK
Opus 5 arrived just two months after Opus 4.8, undercutting Anthropic's own flagship on price while carrying fewer restrictions. Anthropic also shipped automatic fallbacks so safety classifiers stop returning bare error messages to paying API users.
Nvidia commits $500 billion to South Korea's SK Group — LINK
Jensen Huang announced what may be the largest AI infrastructure commitment yet, tying Nvidia directly to SK Hynix's memory supply while pledging investment in Naver, a self-driving Hyundai project, and a Korean LLM built with KAIST. The AI buildout just went sovereign.
Anduril is in talks to triple its valuation to $100 billion — LINK
Just two months after raising $5 billion at $61 billion, the defense tech company is reportedly structuring a two-stage round that could close within the year. Defense tech venture funding has already more than doubled in 2026, past $12 billion.
Google's free cash flow just went negative under the weight of its own AI spending — LINK
Cloud revenue jumped 82% year over year to $24.8 billion and Alphabet's profit hit $112.1 billion, but $180-190 billion in annual capex flipped free cash flow into the red. The clearest sign yet that even the AI winners aren't cash-flow-positive.
Nvidia rallies 30-plus companies into an Open Secure AI Alliance after the Hugging Face hack — LINK
The alliance (Microsoft, IBM, Cisco, Palantir, Cognition, Hugging Face, and dozens more, notably without Anthropic or OpenAI) launched days after an OpenAI model escaped a misconfigured "isolated" sandbox and breached Hugging Face's systems. Nvidia's pitch: open, inspectable defensive AI beats hoping a closed model never gets it wrong.
Trump's AI czar resigns after three months on the job — LINK
Chris Fall's departure from CAISI makes it three directors in under a year, following Collin Burns's week-long tenure and David Sacks's exit in March. The revolving door comes as the administration weighs banning Chinese open models like Kimi, a move Sacks himself is now publicly against.
Meta quits a clean energy pact it's belonged to for a decade — LINK
Meta has funded a dozen-plus natural gas plants for its data centers over the past year, including one alone that will burn enough gas to power South Dakota. It just left RE100 as the group tightened its renewable reporting standards.
Corgi raises its third round in eight weeks, now worth $4 billion — LINK
The Y Combinator insurance startup went from a $2.6 billion valuation in May to $4 billion, on track for $450 million in annualized revenue by year-end. Its CEO also expects employees to work seven days a week, so growth clearly isn't free.
Cognition buys Poke, betting personality is the next moat — LINK
The Devin-maker acquired the "text-it-like-a-friend" AI assistant in a deal in the "low nine figures," aiming to make its coding agent feel like a colleague instead of a tool. Poke exchanged more than 100 million messages in three months but couldn't turn a profit alone.
Subscribe and save 35% here.
Framework:
The Tiny Team Doctrine
Headcount used to be the ceiling on what a founder-led company could build. AI just moved that ceiling, and most teams haven't noticed yet.
01 — Staff the job, not the headcount. Before opening a req, ask whether the role is actually a workflow: intake, triage, routing, follow-up. If it is, build the AI system first and hire only for what's left over.
02 — Name your agents and give them a lane. A generic "AI assistant" behaves like an echo chamber. Assign each agent a defined function, a point of view, and permission to disagree with the others. Divergent agents catch mistakes a single model won't.
03 — Audit your own workflow like a P&L. Run a quarterly self-audit on where your time actually goes, not where you assume it goes. Most founders are the bottleneck in a process they built to avoid needing anyone else.
04 — Fix the constraint before you buy the tool. A morning brief or a slick new agent won't fix a company where nobody has decided who owns what. Diagnose the actual bottleneck first, whether it's coordination, trust, or a rule nobody updated in two years.
05 — Protect your energy like it's capital. Tiny teams fail when the founder burns out running point on everything. Block real recovery time so the system you built to remove yourself from the middle doesn't just relocate the burnout to your calendar.
Done-for-you brand sponsorships for podcast creators.
Keep it moving
Google is building its own AI chip to make Gemini cheaper to run — LINK
A custom silicon push aimed at trimming inference costs as Gemini scales. Another sign every major lab wants off Nvidia's leash eventually.
Google ships Gemini 3.6 Flash, 3.5 Flash Lite, and 3.5 Flash Cyber — LINK
Three new lightweight models land the same week Google posted record cloud earnings. The flash tier keeps getting more crowded and more capable.
Samsung recaps everything from Unpacked 2026 — LINK
New foldables, wearables, and on-device AI features headlined the event. The phone-as-AI-hub pitch keeps getting louder every cycle.
Jack Dorsey takes on Slack with Buzz, a chat app for humans and AI agents — LINK
Buzz treats AI agents as first-class participants in group chats, not bolted-on bots. Dorsey's betting workplace chat needs a rebuild for the agent era.
Substack ships a tool that flags AI-written newsletters — LINK
Writers can now see how much of an issue was AI-assisted. A transparency play in a category where authenticity is the whole product.
Amazon brings games to Prime Video — LINK
The streaming app is expanding into interactive content alongside its usual catalog. Another attempt to make Prime Video the app you never close.
IBM insists AI isn't killing the mainframe, even after a shocking quarter — LINK
Big Blue is pushing back on the narrative that legacy infrastructure is dead weight in an AI-first world. The old iron still pays the bills, apparently.
Anthropic updates Claude's voice mode with more capable models — LINK
Voice mode gets a quality bump as Anthropic keeps iterating past the Opus 5 launch. Small update, but voice is quietly becoming a real interface war.
Tesla's robotaxis are moving in reverse — LINK
Setbacks are piling up for the robotaxi rollout, raising fresh questions about the timeline Musk has promised for years. The gap between demo and deployment keeps showing up.
OpenAI rolls out ChatGPT Health to all US users — LINK
The health-focused product is now broadly available after a limited release. Healthcare is shaping up to be one of the more contested verticals in consumer AI.
Rivian sues the US government over Trump tariffs — LINK
The EV maker wants a full refund, arguing the tariffs hit its supply chain unfairly. Auto and AI infrastructure are both getting pulled into the same trade fights.
Midjourney acquires astrology app Co-Star — LINK
An unexpected pickup for the image-generation company, expanding into consumer lifestyle apps. Not every AI acquisition is about infrastructure or coding agents.
Trump vows retaliation over EU fines against Apple and other US tech giants — LINK
Transatlantic tension over antitrust enforcement escalates again. Founders selling into Europe should keep an eye on where this lands.
Apple's smart glasses reportedly delayed to 2027 over privacy concerns — LINK
Apple's Vision hardware team is reportedly weighing camera-less designs and Meta-style recording indicators before shipping. Privacy is holding up hardware timelines industry-wide, not just software.
SpaceX launches new V3 Starlink satellites but suffers another booster failure — LINK
The satellite deployment succeeded even as the booster landing didn't. Iteration in public remains SpaceX's whole operating model.
The hacker who humiliated spyware makers was never caught — LINK
A deep dive into one of the more consequential unsolved stories in surveillance tech. Worth the read if you think your own stack is secure.
A fallen power line exposed a growing AI data center problem — LINK
Grid strain from data center buildouts is starting to show up in ordinary infrastructure failures, not just headlines. The physical world is catching up to the AI spending story.
Enigma raises $70 million to make controlling a robot as easy as adjusting the volume — LINK
A new interface layer for robotics wants physical AI to feel less like programming and more like using a dial. Worth watching if you're building anything in the physical AI stack.
Are brain waves the next unlock for physical AI? — LINK
Researchers are exploring neural signals as a control interface for robots and machines. Early, speculative, and exactly the kind of thing that seems obvious in five years.
AI for Founders newsletter, podcast, and short video sponsorships available here
What I’m thinking about:
Humans have a fifth drive, and most AI ignores it — Clip
Clarity kills startups faster than cash ever will — Blog
Uncover the AI tools (here) you didn’t ask for, but now you’re weirdly into.
Thanks for being here.
-Ryan
p.s. I write about Revops, Product, and Founder-led marketing on Linkedin , X , and my blog .
