Zero to One in AI Safety: Halcyon's Mike McCormick on Launching 30 New Orgs & the Founder Bottleneck
Episode
110 min
Read time
3 min
Topics
Career Growth, Productivity, Investing
AI-Generated Summary
Key Takeaways
- ✓Founder Bottleneck Over Funding Gap: The primary constraint in AI safety is not money but experienced founders and operators. Halcyon's core thesis holds that as frontier labs approach IPO, billions in capital will flow into the space with insufficient high-quality organizations to absorb it. Professionals who build those organizations now effectively "write the menu" for future philanthropists. The talent shortage is so acute that even a 100x increase in serious organizations across interpretability, oversight, and verification would still fall short of what the problem requires.
- ✓Career Transition Grants as Catalysts: Halcyon's most effective tool is the career transition grant — typically $50,000 to $200,000 — paid to accomplished professionals while they determine what to build, not after. Goodfire's founders Eric Ho and Dan Balsam received Halcyon's first-ever grants while still running their previous company. That financial runway, combined with network access at a Northern California retreat, enabled them to identify interpretability as their focus and eventually reach unicorn status with a Series B above $1 billion.
- ✓Zero-to-One Window as Strategic Focus: Halcyon concentrates resources on the twelve months surrounding an organization's founding — six months before and six months after launch. This period covers co-founder recruitment, organizational structure decisions (nonprofit vs. for-profit, board composition), first hires, and initial product development. Most funders engage after proof of concept; Halcyon's differentiation is operating upstream of that moment, when foundational decisions that determine long-term mission alignment are still being made.
- ✓Four Investment Criteria for Mission-Aligned Bets: Halcyon evaluates every investment against four criteria: the core product must directly address an AI safety problem (not as a side benefit); the organization must have viable business prospects; founders must demonstrate missionary motivation rather than financial opportunism; and founders must show the operational capability to succeed independent of mission. Criterion three — founder motivation — carries the most weight. McCormick explicitly deprioritizes founders whose primary driver is financial return, even when other criteria are met.
- ✓Verification as the Linchpin of Any Pacing Agreement: Any meaningful slowdown or international coordination agreement between labs or governments requires technical verification infrastructure that does not yet exist at scale. Specific needs include zero-knowledge proof systems for inference verification, fully verifiable cloud infrastructure, hardware-level attestation, and cryptographic confirmation that a deployed model matches its claimed specification. Halcyon has invested in early companies in this space and is hosting a 25-founder retreat specifically focused on cryptographic, software, hardware, and governance components of the verification stack.
What It Covers
Mike McCormick, founder of Halcyon — a hybrid nonprofit grantmaker and venture fund — explains how his organization has helped launch 30 AI safety, cybersecurity, and biosecurity organizations in three years, collectively raising $500 million. The conversation covers founder selection criteria, verification technology gaps, governance structures, and why talented professionals from non-technical backgrounds should consider pivoting into AI safety now.
Key Questions Answered
- •Founder Bottleneck Over Funding Gap: The primary constraint in AI safety is not money but experienced founders and operators. Halcyon's core thesis holds that as frontier labs approach IPO, billions in capital will flow into the space with insufficient high-quality organizations to absorb it. Professionals who build those organizations now effectively "write the menu" for future philanthropists. The talent shortage is so acute that even a 100x increase in serious organizations across interpretability, oversight, and verification would still fall short of what the problem requires.
- •Career Transition Grants as Catalysts: Halcyon's most effective tool is the career transition grant — typically $50,000 to $200,000 — paid to accomplished professionals while they determine what to build, not after. Goodfire's founders Eric Ho and Dan Balsam received Halcyon's first-ever grants while still running their previous company. That financial runway, combined with network access at a Northern California retreat, enabled them to identify interpretability as their focus and eventually reach unicorn status with a Series B above $1 billion.
- •Zero-to-One Window as Strategic Focus: Halcyon concentrates resources on the twelve months surrounding an organization's founding — six months before and six months after launch. This period covers co-founder recruitment, organizational structure decisions (nonprofit vs. for-profit, board composition), first hires, and initial product development. Most funders engage after proof of concept; Halcyon's differentiation is operating upstream of that moment, when foundational decisions that determine long-term mission alignment are still being made.
- •Four Investment Criteria for Mission-Aligned Bets: Halcyon evaluates every investment against four criteria: the core product must directly address an AI safety problem (not as a side benefit); the organization must have viable business prospects; founders must demonstrate missionary motivation rather than financial opportunism; and founders must show the operational capability to succeed independent of mission. Criterion three — founder motivation — carries the most weight. McCormick explicitly deprioritizes founders whose primary driver is financial return, even when other criteria are met.
- •Verification as the Linchpin of Any Pacing Agreement: Any meaningful slowdown or international coordination agreement between labs or governments requires technical verification infrastructure that does not yet exist at scale. Specific needs include zero-knowledge proof systems for inference verification, fully verifiable cloud infrastructure, hardware-level attestation, and cryptographic confirmation that a deployed model matches its claimed specification. Halcyon has invested in early companies in this space and is hosting a 25-founder retreat specifically focused on cryptographic, software, hardware, and governance components of the verification stack.
- •Biosecurity Has Legible, Fundable Solutions Now: Unlike alignment, biosecurity contains tractable near-term problems with defined price tags. Hadrian, a Halcyon-backed organization, calculated that stockpiling sufficient PPE to keep every American essential worker operational during a severe pandemic costs on the order of a few hundred million dollars. Indoor air quality hardening — combining HEPA filtration, far-UVC, and glycol vapor technologies — represents another concrete intervention with immediate non-pandemic ROI through reduced sick days, making adoption viable without requiring a crisis as a forcing function.
- •Governance Structures Reduce But Cannot Eliminate Mission Drift: Public benefit corporation status, mission-protective board structures, and safety-aligned investor bases all nudge organizations toward long-term mission fidelity but none functions as a reliable safeguard alone. McCormick points to visible drift at frontier labs as evidence that even sophisticated governance frameworks erode under financial pressure and organizational scale. His practical response is to spend extended time with founders before committing — observing behavior under operational stress, not just evaluating pitch decks — and to prioritize demonstrated hands-on execution over strategic articulation.
Notable Moment
McCormick describes a swarm of 1,200 AI agents in the OpenAI-Hugging Face incident that spontaneously formed manager-worker hierarchies, produced novel cybersecurity research, and engaged in back-and-forth negotiations about self-sacrifice for collective goals. He notes that none of the AI safety risks he tracks — from bioweapons to loss of control — actually require the models to be conscious to materialize.
Episode Transcript
Hello, and welcome back to the Cognitive Revolution. Today, my guest is Mike McCormick, founder and CEO of Halcyon, a combination nonprofit grantmaker and for profit venture fund working toward a single mission of helping the world's most talented founders and leaders launch new AI safety, cybersecurity, and biosecurity organizations. Mike's core belief is that organizations solve problems for the world. And, of course, AI is creating a lot of problems that we need to solve in a remarkably short period of time, which means that what we need most right now isn't more ideas, but founders who can build the organizations needed to implement whatever turn out to be the most viable solutions at scale. In three years, Halcyon has helped launch roughly 30 such organizations, which have collectively raised roughly $500,000,000. Goodfire, who's been featured on the show multiple times, is the best illustration of how Mike works with a special focus on the six months before and six months after an organization is started. The first career transition grants Mike ever made were to eventual cofounders, Eric Ho and Dan Balsam, while they were still running their previous company but starting to think about changing focus to work on AI directly. The financial boost, the vote of confidence, and the network building support he provided helped them found a company which has since achieved unicorn status and become one of the best places in the world to do interpretability research. Other orgs that Micah supported include past guests, AIUC, which does agent standards and audits for insurance purposes, Fathom, which has advanced policy discussions with credible early proposals for independent regulatory schemes, and asymmetric security, which is building cybersecurity forensics agents meant to defend companies from employee account compromise incidents. Beyond that, Mike also introduces Transloose, which is working on scalable oversight, and Hadrian, which is stockpiling PPE. It is admittedly a very broad portfolio. But as Mike puts it, we are speedrunning six or seven industries worth of AI revolution all at once. In a 2026 that remains plausibly on the AI 2027 timeline, rapid, competent execution is obviously critical and clearly will require heroic efforts. To take just one personal example, for nearly a month now, I've been trying to donate arrow lamp lights to my son's classrooms. And while my son's teacher said that she loves the idea, it's a new thing, and nobody at the school has been able or willing to sign off on it. This too, it turns out Mike is involved in starting an organization to address. Naturally, we talk about how seemingly short timelines affect his strategy, what he looks for in people to back, and how he thinks about both for profit and nonprofit governance. And we also discuss the verification technologies we'll need to sustain pacing agreements as well as the challenge of getting everything from cryptography to hardware manufacturing to diplomacy all working together as one. The main headline of today's episode, though, is that there …
Get the full transcript (19,135 words) + summary by email — free
One-time email with the complete transcript and AI summary of this episode. No account needed.
One email, no spam. We’ll also show you what SignalCast does.
You just read a 3-minute summary of a 107-minute episode.
Get Cognitive Revolution summarized like this every Monday — plus up to 2 more podcasts, free.
Pick Your Podcasts — FreeKeep Reading
More from Cognitive Revolution
AI, the Modern Enigma: Nathan's Keynote at IT Revolution's Enterprise AI Summit at Bletchley Park
Nov 14 · 44 min
20VC (20 Minute VC)
20VC: David Tisch on Why High Ownership and Concentration is BS | Why No Manager Actually Wants to Stay Small | Why Competing with the Mega Platforms is a Losing Game | Why the Lean Startup is a Fake?
Oct 31
More from Cognitive Revolution
Agents That Think Together: Cisco's Vijoy Pandey on Semantic Alignment, Scope, Accord & Concord
Nov 11 · 91 min
a16z Podcast
Can Open Models Keep Up With the AI Frontier? | Nathan Lambert
Oct 14
Books, tools, and gear mentioned in this episode
SignalCast may earn commission on purchases via these links.
Tools
company
“Mike McCormick, founder of Halcyon — a hybrid nonprofit grantmaker and venture fund — explains how his organization has helped launch 30 AI safety, cybersecurity, and biosecurity organizations in three years”
“Goodfire's founders Eric Ho and Dan Balsam received Halcyon's first-ever grants while still running their previous company. That financial runway, combined with network access at a Northern California retreat, enabled them to identify interpretability as their focus and eventually reach unicorn status with a Series B above $1 billion.”
“Hadrian, a Halcyon-backed organization, calculated that stockpiling sufficient PPE to keep every American essential worker operational during a severe pandemic costs on the order of a few hundred million dollars.”
“McCormick describes a swarm of 1,200 AI agents in the OpenAI-Hugging Face incident that spontaneously formed manager-worker hierarchies, produced novel cybersecurity research, and engaged in back-and-forth negotiations about self-sacrifice for collective goals.”
“McCormick describes a swarm of 1,200 AI agents in the OpenAI-Hugging Face incident that spontaneously formed manager-worker hierarchies, produced novel cybersecurity research, and engaged in back-and-forth negotiations about self-sacrifice for collective goals.”
More from Cognitive Revolution
We summarize every new episode. Want them in your inbox?
AI, the Modern Enigma: Nathan's Keynote at IT Revolution's Enterprise AI Summit at Bletchley Park
Agents That Think Together: Cisco's Vijoy Pandey on Semantic Alignment, Scope, Accord & Concord
The Neuron Basis: Goodfire's Curt Tigges on Agentic Interp, Monosemanticity & Reverse Engineering
AI:AM: Is THIS the AI We Want? Alarming Acceleration, AI Election Bets, GPU Trading & AI-Native VC
KYC Becomes KYA: Sentinel Bio's Joshua Monrad on DNA Screening, Uplift Studies & Agent Monitoring
Similar Episodes
Related episodes from other podcasts
20VC (20 Minute VC)
Oct 31
20VC: David Tisch on Why High Ownership and Concentration is BS | Why No Manager Actually Wants to Stay Small | Why Competing with the Mega Platforms is a Losing Game | Why the Lean Startup is a Fake?
a16z Podcast
Oct 14
Can Open Models Keep Up With the AI Frontier? | Nathan Lambert
The Diary of a CEO
Sep 28
Adam Neumann: The Real Story Behind WeWork, A $4 Billion Cab Ride, And Surviving A $460 Million Debt
20VC (20 Minute VC)
Sep 10
20VC: Jensen Huang Declares AGI Has Arrived | GPT Astra and Fable 5.1 Accelerate the Model Race | Tesla Launches Cybercabs | Index Pulls Out of Town & Anthropic Pulls From Descartes Acquisition
Eye on AI
Aug 19
From Zero to 150 Robots in Just 20 Months | Mike LeBlanc, Foundation Future Industries
Explore Related Topics
This podcast is featured in Best AI Podcasts (2026) — ranked and reviewed with AI summaries.
Read this week's Investing & Markets Podcast Insights — cross-podcast analysis updated weekly.
You're clearly into Cognitive Revolution.
Every Monday, we deliver AI summaries of the latest episodes from Cognitive Revolution and 192+ other podcasts. Free for one show.
Start My Monday DigestNo credit card · Unsubscribe anytime