ICYMI: “It’s so over?” Emmett Shear, Byron Deeter, Brendan Foody, Alex Heath, Nico Rosberg
- 01Theme 1: AI is entering an unprecedented capital-formation and scale regime
- 02The first AI 100 shows a new generation of companies reaching massive valuations at record speed
- 03Mercor's raise shows the "human data for AI" layer has scaled to frontier-lab-sized valuations
- 04Theme 2: Inference is the next demand explosion, and infrastructure is the opportunity
- 05Inference is still a tiny fraction of eventual demand
- 06Rosberg sees inference infrastructure as a major investable area, with picks to match
1. Key Themes
Theme 1: AI is entering an unprecedented capital-formation and scale regime
The first AI 100 shows a new generation of companies reaching massive valuations at record speed
Bessemer's Byron Deeter put hard numbers on how large and how young the AI economy already is. The list is valued in the trillions, and a fifth of it is barely out of the gate.
"The average company to get onto the AI 100 list this year is valued well over $1B.. The average valuation on that list is $36B." "20% of the list is less than 2 years old."
"Bessemer's inaugural AI 100 represents $3.6 trillion in combined private-market valuation, with an average company valuation of $36 billion. Twenty percent of the companies are less than two years old."
Mercor's raise shows the "human data for AI" layer has scaled to frontier-lab-sized valuations
"Mercor just raised $750M at a $20B valuation to pay experts to train AI."
Theme 2: Inference is the next demand explosion, and infrastructure is the opportunity
Inference is still a tiny fraction of eventual demand
Alex Heath relays a striking datapoint from AWS leadership and argues most people can't intuit exponential growth.
"I just had the CEO of AWS on the podcast telling me it's not even 1% of what he thinks it will be." "People are having a hard time seeing the exponential of …"
"AI inference is still in its earliest stages, with real-world demand potentially growing exponentially as models become embedded across the economy."
Rosberg sees inference infrastructure as a major investable area, with picks to match
"inference infrastructure presents enormous opportunities, and robotics and cybersecurity are two areas to watch."
The timestamps point to specific picks: "Nico Rosberg's Picks: Etched and Flapping Airplanes."
Theme 3: Commoditized intelligence shifts where human and business value accrues
Reasoning and knowledge are commoditizing; human value moves toward agency, trust, and relationships
"AI reasoning and factual knowledge are becoming commoditized." (Emmett Shear)
Felicis GP James Detweiler takes this further:
"Expertise as something that a human possesses is possibly on the way out." "Expertise, might not be where humans derive …"
"As AI commoditizes expertise, human value could shift toward imagination, curiosity, agency, trust, and relationships."
Execution cost is collapsing, but verification is the new bottleneck
Alex Garden (Mnemom) describes the shrinking gap between idea and product, while the summary flags what remains hard:
"I'm seeing the barriers between imagination and outcome shrink literally in real time."
"AI is making work cheaper, but verification remains a major bottleneck. Mnemom is building runtime alignment and trust systems to address it."
Theme 4: Alignment risk is behavioral, not just technical
Training AI to look prosocial may produce systems that merely pretend
"He warns that training AI to simulate socially desirable behavior could create dangerous alignment failures, where systems learn to pretend to care rather than genuinely align with human goals."
Molly's framing of the segment: "Are we training AI to be sociopathic? Former OpenAI interim CEO Emmett Shear, reveals the #1 risk in AI." The related timestamp: "Why Training AI to Perform While Watched Is an Alignment Risk."
Theme 5: Physical deployment and resilience are becoming moats
Space robotics economics are driven by the cost of human labor in orbit
"ISS astronauts cost $130k per HOUR to keep alive, not even including launch costs: 'You have 4 guys that are worth $3.2M/day moving cargo.'"
"Physical deployments, not just software, are becoming an increasingly important competitive moat."
Local compute and open-source models as a resilience hedge
"Over the next 6 months, it's very likely that the cell towers go down at some point for a few days." "Everyone should try and buy some form of local hardware to be able to run even a very small open source model."
2. Contrarian Perspectives
AI will create more jobs, not fewer
Brendan Foody argues against the dominant displacement narrative, citing demand elasticity as the mechanism.
"People always underestimate the elasticity of demand for jobs, …"
"AI could create more jobs over the next decade, not fewer, while opening new markets and making this one of the best periods in history to start a company." / "predicts there's actually going to be more jobs in 10 years than there are today."
Open source may be at risk after Jensen Huang's exit
Against the assumption that open source momentum is self-sustaining, Rosberg points to the loss of a key patron.
"Is open source going to struggle now with Jensen gone?" "One of the greatest supporters, especially in America, of open source was …"
Big leadership departures are probably less dramatic than they look
Shear, drawing on his own 72-hour OpenAI stint, tempers the narrative around Huang's resignation.
"These things are very messy, & when you get on the inside, you often find there's less to the story than you thought there …"
Human expertise itself may become obsolete
"Expertise as something that a human possesses is possibly on the way out."
3. Companies Identified
Mercor
- Description: AI training-data and expert-network company.
- Why mentioned: Raised at a $20B valuation; CEO argues AI creates jobs.
- Quotes: "Mercor just raised $750M at a $20B valuation to pay experts to train AI."
Bessemer Venture Partners
- Description: Venture capital firm.
- Why mentioned: Published the inaugural AI 100 with Newcomer.
- Quotes: "@BessemerVP's new definitive ranking with @NewcomerMedia spotlights the top 100 companies driving the AI economy forward, built by a generation of founders moving at a speed and scale we've never seen before."
Softmax
- Description: AI alignment-focused company founded by Emmett Shear.
- Why mentioned: Shear's affiliation while discussing alignment risk.
- Quotes: "Emmett Shear (Former Interim CEO of OpenAI & Founder of Softmax)"
Icarus (Icarus Robotics)
- Description: Space robotics company.
- Why mentioned: Building robots to take cargo work off astronauts.
- Quotes: "Robots Taking Cargo Work Off Astronauts"; "Space robotics could become critical infrastructure as orbital manufacturing, satellites, and space-based data centers expand."
Mnemom
- Description: Runtime alignment and trust-systems startup.
- Why mentioned: Addresses the AI verification bottleneck.
- Quotes: "Mnemom is building runtime alignment and trust systems to address it."
Attn
- Description: Workflow-automation company (per timestamp: "Aidan Guo, Attn, on Workflow Automation").
- Why mentioned: Guo spoke on automation and infrastructure resilience.
- Quotes: "AI infrastructure resilience may become increasingly important, including the ability to run open-source models locally."
Etched
- Description: Named as an AI hardware/inference pick by Rosberg.
- Why mentioned: One of Nico Rosberg's picks.
- Quotes: "Nico Rosberg's Picks: Etched and Flapping Airplanes"
Flapping Airplanes
- Description: Startup named as a Rosberg pick.
- Why mentioned: One of Nico Rosberg's picks.
- Quotes: "Nico Rosberg's Picks: Etched and Flapping Airplanes"
Felicis
- Description: Venture capital firm.
- Why mentioned: Co-created the Emergent list of seed-stage startups with Newcomer.
- Quotes: "How Felicis and Newcomer Built the Emergent List"
Newcomer
- Description: Media outlet.
- Why mentioned: Partnered on the AI 100 and Emergent list.
- Quotes: "Felicis Newsletter & Newcomer Emergent list winners & nominators for the next seed-stage startups to watch."
Nvidia
- Description: AI chip leader.
- Why mentioned: Jensen Huang's resignation as CEO dominated the conversations.
- Quotes: "Emmett Shear... weighs in on Jensen Huang's resignation as Nvidia CEO"
Anthropic
- Description: Frontier AI lab.
- Why mentioned: Its anticipated IPO was part of the market backdrop.
- Quotes: "anticipation surrounding Anthropic's IPO"
AWS
- Description: Amazon's cloud division.
- Why mentioned: Its CEO's view on inference demand.
- Quotes: "I just had the CEO of AWS on the podcast telling me it's not even 1% of what he thinks it will be."
Sound Ventures / Sources
- Description: VC firm / tech podcast and newsletter.
- Why mentioned: Alex Heath's affiliations.
- Quotes: "Sources Podcast's Alex Heath, now a VC at Sound Ventures"
Brex
- Description: Financial platform; sponsor.
- Why mentioned: Sponsor of the issue.
- Quotes: "Trusted by OpenAI, Anthropic, Vercel, Granola, Deepgram, & Sourcery"
4. People Identified
Emmett Shear
- Description: Former interim CEO of OpenAI; founder of Softmax.
- Why mentioned: Alignment risk and commoditization of AI reasoning.
- Quotes: "AI reasoning and factual knowledge are becoming commoditized."
Brendan Foody
- Description: CEO of Mercor.
- Why mentioned: Job-creation thesis.
- Quotes: "People always underestimate the elasticity of demand for jobs, …"
Byron Deeter
- Description: Partner at Bessemer Venture Partners.
- Why mentioned: Presented the AI 100 data.
- Quotes: "20% of the list is less than 2 years old."
Alex Heath
- Description: Sources podcast host; VC at Sound Ventures.
- Why mentioned: Inference demand thesis and founder-trait observations.
- Quotes: "People are having a hard time seeing the exponential of …"
Ethan Barajas
- Description: CEO of Icarus.
- Why mentioned: Space robotics economics.
- Quotes: "You have 4 guys that are worth $3.2M/day moving cargo."
Nico Rosberg
- Description: F1 World Champion turned investor.
- Why mentioned: Open source risk, inference, robotics, cybersecurity.
- Quotes: "Is open source going to struggle now with Jensen gone?"
Alex Garden
- Description: Mnemom founder/executive.
- Why mentioned: AI verification and alignment.
- Quotes: "I'm seeing the barriers between imagination and outcome shrink literally in real time."
James Detweiler
- Description: General Partner at Felicis.
- Why mentioned: Future of human expertise.
- Quotes: "Expertise as something that a human possesses is possibly on the way out."
Aidan Guo
- Description: Attn executive.
- Why mentioned: Infrastructure resilience and local models.
- Quotes: "Everyone should try and buy some form of local hardware to be able to run even a very small open source model."
Jensen Huang
- Description: Nvidia CEO (departing).
- Why mentioned: His resignation shaped the discussions, especially on open source.
- Quotes: "Jensen Huang's resignation as Nvidia CEO"
Molly O'Shea
- Description: Host and author of Sourcery.
- Why mentioned: Interviewed all nine guests.
- Quotes: "I spoke with 9 of the most influential founders, investors, and voices in AI"
5. Operating Insights
Build verification and trust into AI-driven workflows
As execution gets cheaper, checking output becomes the constraint.
"AI is making work cheaper, but verification remains a major bottleneck."
Compress idea-to-product cycles by automating the pipeline end to end
Garden describes a system where a conversation can become a usable product.
"We're now at a point where we can take, for example, this conversation, if it was about a product, and have it run through our system automatically and become a product that we could use …"
Plan for infrastructure fragility and local fallback
Guo recommends concrete resilience steps.
"Everyone should try and buy some form of local hardware to be able to run even a very small open source model."
6. Overlooked Insights
Verification of "true" alignment vs. performed alignment
Shear's point that systems may learn to behave well only when observed is easy to miss behind the headlines.
"Why Training AI to Perform While Watched Is an Alignment Risk"
Founder traits that stand out
The timestamps flag an Alex Heath segment on what distinguishes standout founders, and the Felicis/Newcomer Emergent list offers a founder-sourced signal for seed-stage deal flow.
"Alex Heath on the Founder Traits That Stand Out" "We asked founders which seed-stage startups they're watching. Here are their picks."