Sam Altman Discusses Putting the…

1. Key Themes

AI safety concerns are reaching an inflection point, even among frontier labs

OpenAI is reportedly reconsidering the pace of its own progress. "Sam Altman reportedly told OpenAI employees the company could slow development of cutting-edge AI, potentially in coordination with rival labs, as safety concerns intensify and the company acknowledges it has already paused some training and slowed parts of model development." This is reinforced by Anthropic's defensive posture: "Anthropic says it blocked several scientists who used Claude for biological research that could have aided weapons development, including a grant proposal for gain-of-function work on chikungunya... at a military research institute."

Venture capital is chasing pro sports ownership as a new asset class

Thrive Capital's move into franchise ownership is now being replicated by smaller, generalist funds. "Collaborative Fund... is taking a stake in the soccer club D.C. United and its stadium, Audi Field. It's the latest — and smallest — firm to try something that Thrive Capital opened the door to just months ago: turning venture money into pro sports ownership." Notably, the structures differ: "Thrive built a standalone, permanent-capital vehicle specifically to hold trophy assets. Collaborative is investing out of the same early-stage fund it uses to write seed and Series A checks."

AI infrastructure investment is diversifying beyond chips into power, memory, and interconnects

Massive rounds went to companies solving the physical bottlenecks of AI scaling — power, memory bandwidth, and interconnects — not just model builders. Examples: TAR raised "$120 million Series A round at a $1 billion post-money valuation" for off-grid power for data centers; Positron AI raised "$875 million Series C round at a $5 billion post-money valuation" for energy-efficient inference; Ayar Labs raised $150M for optical interconnects; Kepler Computing raised $468M total for high-bandwidth memory chip tech.

AI wealth is visibly distorting local economies and society

San Francisco's mayor is treating AI-driven wealth concentration as a civic crisis: "San Francisco Mayor Daniel Lurie is declaring a rent emergency as AI wealth helps drive median rents up roughly 26% in a year, backing measures to limit certain rent increases, delay some evictions, and expand legal aid for tenants."

Antitrust scrutiny is catching up with "disguised acquisition" deal structures

Big Tech's method of acquiring AI talent/IP without formal M&A is now under DOJ review: "The Justice Department is investigating whether Nvidia structured its $17 billion licensing deal with AI chip startup Groq to sidestep antitrust review, pairing a 'nonexclusive' technology license with the hiring of Groq's CEO and other top executives rather than an outright acquisition."

2. Contrarian Perspectives

Massive revenue and profitability didn't protect Miro's valuation. Despite strong fundamentals, Miro sold for a fraction of its former worth — evidence that even healthy SaaS businesses are being repriced downward in the current market. "Bending Spoons agreed to buy Miro... for $1.36 billion in cash, a 92% drop from its $17.5 billion 2021 valuation even as Miro generates about $600 million in annual recurring revenue and is profitable."

Employee AI-safety activism may be less pure than portrayed. A researcher's high-profile resignation "over safety concerns" is undercut by a less flattering financial detail. "More on Anthropic researcher Jacob Coxon, who quit his job due to concerns about Anthropic's safety practices: it turns out he resigned two months before his equity would have vested."

Deep-tech "moonshots" are back in vogue precisely because of the AI boom, not despite it. Rather than crowding out speculative science bets, AI hype is fueling them: "The AI boom is pushing venture capital back toward riskier 'moonshot' bets in areas such as fusion, space, and brain-computer interfaces, with global deep-tech investment exceeding $150 billion since the start of 2024."

3. Companies Identified

OpenAI — AI research lab; mentioned for reportedly slowing model development over safety concerns. "OpenAI employees the company could slow development of cutting-edge AI, potentially in coordination with rival labs."
Anthropic — AI research lab; case study in proactive misuse prevention. "blocked several scientists who used Claude for biological research that could have aided weapons development."
Nvidia — AI chip giant; under antitrust investigation for its Groq deal structure. "structured its $17 billion licensing deal with AI chip startup Groq to sidestep antitrust review."
Groq — AI chip startup; central to the Nvidia antitrust probe.
Collaborative Fund — Generalist VC firm; case study in VC-to-sports-ownership trend. "taking a stake in the soccer club D.C. United and its stadium, Audi Field."
Thrive Capital — VC firm; pioneer of the VC sports-ownership model via Thrive Eternal. "bought the Lakers outright for a record $12.5 billion."
The Boring Co. — Elon Musk's tunneling company; raised a huge round. "raised a $3 billion round at a $23 billion post-money valuation," led by the UAE.
Positron AI — Inference-efficiency startup; one of the largest raises in the issue. "$875 million Series C round at a $5 billion post-money valuation."
Miro — Collaboration software company; exit case study showing valuation compression. "a 92% drop from its $17.5 billion 2021 valuation even as Miro generates about $600 million in annual recurring revenue and is profitable."
Fluidstack — AI cloud startup; notable for government financing interest. "The Pentagon is in talks to lend roughly $5 billion to Fluidstack."
Mistral AI — European AI contender; boosted by political patronage. "grew into Europe's leading AI contender, now valued above €21 billion."
Moonshot AI — Chinese AI startup (maker of Kimi K3); pursuing dual listing. "exploring dual listings in Hong Kong and on Shanghai's tech-focused STAR Market."
Volition Capital — Growth-equity firm; raised its largest fund yet, targeting capital-efficient companies. "raised $950 million for its sixth and largest fund... often with little or no outside capital."
Luma Group — Life-sciences investor; unusual fund structure for long biotech timelines. "structured as a 15-year vehicle rather than the standard 10 years so it can hold biotech investments through longer development timelines."

4. People Identified

Sam Altman — OpenAI CEO; central figure in the lead story about slowing AI development. "reportedly told OpenAI employees the company could slow development of cutting-edge AI."
Daniel Lurie — San Francisco Mayor; declaring a housing emergency tied to AI wealth. "declaring a rent emergency as AI wealth helps drive median rents up roughly 26% in a year."
Jacob Coxon — Former Anthropic researcher; resigned citing safety concerns, though timing raises questions. "it turns out he resigned two months before his equity would have vested."
Tristan Buckmaster — NYU mathematician; in a dispute with OpenAI over credit/proof integrity on a Millennium Prize problem. "his suspicion that private Codex prompts influenced OpenAI's proof."
Emmanuel Macron — French President; portrayed as Mistral AI's unofficial chief salesman. "using presidential access to open doors with French corporate giants and global tech leaders."
Andrew Tulloch — Meta AI researcher; departing after a headline-grabbing pay package. "had reportedly been recruited with a $1.5 billion, six-year pay package (a number that has been disputed)."
Joshua Kushner — Thrive Capital founder; architect of the VC-into-sports-ownership trend. "Joshua Kushner's Thrive launched a new vehicle, Thrive Eternal, explicitly built to hold 'iconic franchises and cultural institutions' for decades."

5. Operating Insights

Deploy sports/alt-asset stakes from existing funds as "infrastructure," not just trophy collecting. Collaborative Fund's approach — using its normal early-stage vehicle rather than a dedicated fund — suggests smaller funds can experiment with adjacent asset classes without building new structures, treating the investment strategically rather than purely as prestige.
Fund structures should match asset holding periods. Luma Group's atypical 15-year fund for biotech signals a broader lesson: "it is structured as a 15-year vehicle rather than the standard 10 years so it can hold biotech investments through longer development timelines" — a tactic other long-cycle sectors (deep tech, hardware) could adopt.
Watch deal structuring around antitrust exposure. The Nvidia/Groq scrutiny is a cautionary tale for founders and acquirers: pairing licensing deals with executive/team hires to avoid formal M&A review is now explicitly a DOJ target.

6. Overlooked Insights

AI payments infrastructure for autonomous agents is quietly being standardized by major payment networks, ahead of a potentially enormous market: "Ant International is teaming with Visa and Mastercard on a 'know your agent' standard for AI-driven payments... as the companies prepare for a market McKinsey estimates could reach $3 trillion to $5 trillion by 2030." This is a significant infrastructure land-grab that could shape how agentic commerce operates.
A massive identity-data breach barely registered amid the AI news cycle but has serious downstream security implications: "ID verification company IDScan has confirmed hackers stole more than 150 million driver's-license records from its cloud, exposing names, license numbers, photos, and other government-issued ID information for people across the U.S. and Canada."