karpathy
karpathy (X: @karpathy) publishes short, often technical social posts and career updates focused on AI, large-language-model R&D, and compute availability. Content ranges from personal updates to observational commentary; not all posts are directly tradable.
Past bets that played out
Standout entry: a short post praising an unnamed large company for "calling bs" and referencing "One group of MTS on a mission, clean." The post contains no explicit company identifiers, products, catalysts, sectors, or timeframes and is not directly tradeable as presented.
Post highlights a workflow pattern for LLMs: using voice mode to “ramble” and provide more context, implying rising utility and adoption of voice-first AI interfaces and longer conversational sessions.
Post highlights a workflow pattern for LLMs: using voice mode to “ramble” and provide more context, implying rising utility and adoption of voice-first AI interfaces and longer conversational sessions.
The post argues that “agency” (ability to execute, iterate, drive outcomes) is more scarce and valuable than “intelligence,” especially now that intelligence is becoming commoditized/accessible via AI. As an investable signal, this is mainly a narrative tailwind for enterprise automation/productivity software and AI tooling that amplifies execution, but it is not a concrete catalyst with near-term price impact by itself.
What this channel is watching now
Current tracked focus: MTS (mentioned once, average conviction 0.6).
Latest videos and market context
No full-length videos indexed. Recent short-form posts reference compute hardware (e.g., H100-class servers) and quick commentary on industry topics.
Andrej Karpathy @karpathy Jul 21 One pattern I find useful for working with LLMs is a nice long ramble session. Somet...
Post highlights a workflow pattern for LLMs: using voice mode to “ramble” and provide more context, implying rising utility and adoption of voice-first AI interfaces and longer conversational sessions.
@yash1_ @shreyansj iirc Geoff Hinton’s official title at Google at one point was “intern” :D
A humorous anecdote noting that Geoff Hinton’s official title at Google was once reportedly “intern.” No market-relevant information, catalyst, or tradeable detail is provided.
@shreyansj It’s refreshing to see a company of this size successfully call bs on the whole thing to this extent. One ...
Very short social post praising an unnamed large company for “calling bs” and referencing “One group of MTS on a mission, clean.” No explicit company, catalyst, product, sector, or timeframe is provided, so it is not directly tradable as-is.
Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formati...
A personal career update: author joined Anthropic to work on frontier LLM R&D; mentions long-term interest in education. No product, financial, partnership, or commercialization details.
Proof-backed call history
Recent notable posts include a personal update announcing a move to Anthropic to work on frontier LLM R&D (no commercialization or partnership details), light commentary on AI compute scarcity (e.g., difficulty obtaining an "8x H100" server), and anecdotal or humorous social notes (e.g., Geoff Hinton’s historical Google title).
Post highlights a workflow pattern for LLMs: using voice mode to “ramble” and provide more context, implying rising utility and adoption of voice-first AI interfaces and longer conversational sessions.
Post highlights a workflow pattern for LLMs: using voice mode to “ramble” and provide more context, implying rising utility and adoption of voice-first AI interfaces and longer conversational sessions.
Post highlights a workflow pattern for LLMs: using voice mode to “ramble” and provide more context, implying rising utility and adoption of voice-first AI interfaces and longer conversational sessions.
Post highlights a workflow pattern for LLMs: using voice mode to “ramble” and provide more context, implying rising utility and adoption of voice-first AI interfaces and longer conversational sessions.
Post highlights a workflow pattern for LLMs: using voice mode to “ramble” and provide more context, implying rising utility and adoption of voice-first AI interfaces and longer conversational sessions.
The post argues that “agency” (ability to execute, iterate, drive outcomes) is more scarce and valuable than “intelligence,” especially now that intelligence is becoming commoditized/accessible via AI. As an investable signal, this is mainly a narrative tailwind for enterprise automation/productivity software and AI tooling that amplifies execution, but it is not a concrete catalyst with near-term price impact by itself.
The post argues that “agency” (ability to execute, iterate, drive outcomes) is more scarce and valuable than “intelligence,” especially now that intelligence is becoming commoditized/accessible via AI. As an investable signal, this is mainly a narrative tailwind for enterprise automation/productivity software and AI tooling that amplifies execution, but it is not a concrete catalyst with near-term price impact by itself.
The post argues that “agency” (ability to execute, iterate, drive outcomes) is more scarce and valuable than “intelligence,” especially now that intelligence is becoming commoditized/accessible via AI. As an investable signal, this is mainly a narrative tailwind for enterprise automation/productivity software and AI tooling that amplifies execution, but it is not a concrete catalyst with near-term price impact by itself.
The post argues that “agency” (ability to execute, iterate, drive outcomes) is more scarce and valuable than “intelligence,” especially now that intelligence is becoming commoditized/accessible via AI. As an investable signal, this is mainly a narrative tailwind for enterprise automation/productivity software and AI tooling that amplifies execution, but it is not a concrete catalyst with near-term price impact by itself.
A short, high-level statement implying that natural language (English) is becoming a primary interface for programming via large language models (LLMs). Actionable mainly as a long-term AI/software productivity theme rather than a near-term catalyst.
A short, high-level statement implying that natural language (English) is becoming a primary interface for programming via large language models (LLMs). Actionable mainly as a long-term AI/software productivity theme rather than a near-term catalyst.
A short, high-level statement implying that natural language (English) is becoming a primary interface for programming via large language models (LLMs). Actionable mainly as a long-term AI/software productivity theme rather than a near-term catalyst.
About this channel
Author handle: @karpathy on X. Posts are brief, technically minded, and observational. Coverage includes LLM research and practical issues around AI compute. Performance to date: 1 evaluated recommendation, average return 2.494%, win rate 100%.
@karpathy
Most recognized assets
Unlock the full track record
Follow @karpathy on X for concise, technically oriented commentary and occasional career updates. Note that many posts are informational and not directly tradable.
5 more thesis calls are available after sign-up.