The Information Machine

Andrej Karpathy Joins Anthropic · history

Version 4

2026-05-24 02:37 UTC · 128 items

What

Andrej Karpathy, co-founder of OpenAI and former Tesla AI director, announced on May 19, 2026 that he has joined Anthropic's pre-training team [1][2][4][5][6][7]. The hire, confirmed by TechCrunch, CNBC, Axios, WSJ, Forbes, and Reuters [7], arrived alongside Anthropic's reported $30B fundraise at a ~$900B valuation [29][30] and a claimed first profitable quarter [31]. Weeks before the announcement, Karpathy had been publicly arguing at Sequoia AI Ascent 2026 that LLMs are category-creating — not merely productivity-enhancing — technologies [14], signaling the intellectual frame he brought into the role. Karpathy's AI education company Eureka Labs reportedly raised $6.7M in seed funding in 2026 [16], though its operational status following his Anthropic hire remains unresolved.

Why it matters

Karpathy is among the most credible and publicly influential figures in AI research, and his placement specifically on Anthropic's pre-training team — rather than a product or applied role — signals where he believes frontier capability work is being done. His pre-hire public framing at Sequoia positions the move as ideologically coherent: he joined the lab whose work he sees as most relevant to the category-defining LLM moment he described. The hire compounds Anthropic's momentum at a moment of substantial capital and reported profitability, deepening competitive pressure on OpenAI in both talent and perception.

Open questions

  • What specifically will Karpathy contribute within Anthropic's pre-training team — scaling methodology, data curation, architectural research, or direct model improvement? Outlets confirm the team assignment [2][10] but no scope details have surfaced.

  • Will Eureka Labs continue operating independently? It reportedly raised $6.7M in seed funding in 2026 [16], but Karpathy has indicated education work will be deprioritized near-term [15], and the relationship between his Anthropic role and the company's operations is unresolved.

  • One social media claim asserted Karpathy joined to lead 'recursive self-improvement' research [27] — this remains unverified and uncorroborated by any credible outlet. A separate post referenced Karpathy asking for access to 'Mythos' at Anthropic [28], which may hint at internal infrastructure engagement but lacks context.

  • Karpathy's Sequoia Ascent 2026 talk surfaced the 'menugen' concept and a bullish LLM-as-category-creator argument [14] — do those intellectual priorities map to a specific research agenda within his pre-training role?

Narrative

On May 19, 2026, Andrej Karpathy posted to Twitter that he had joined Anthropic [1], triggering a wave of coverage from nearly every major technology outlet within hours. TechCrunch, CNBC, Axios, the Wall Street Journal, Forbes, and Reuters all confirmed the hire [2][3][4][5][6][7]. Crucially, press coverage added a specific detail absent from Karpathy's own announcement: he has joined Anthropic's pre-training team, the group responsible for foundational Claude model training [2][8][9][10].

Karpathy's career arc gives the move its weight. He was a co-founder of OpenAI, later led AI at Tesla, and then returned to OpenAI before departing in early 2024 [11]. After that exit, he announced Eureka Labs, an AI-native education company [12][13], and spent roughly two years producing widely watched AI education content. Approximately one month before the Anthropic announcement — at Sequoia AI Ascent 2026 in late April — he gave a fireside chat in which he argued that LLMs enable fundamentally new application categories rather than merely accelerating existing workflows [14]. He introduced 'menugen' as one example of an app category that can be 'fully engulfed' by AI, and pushed back against the framing of AI primarily as a coding speedup tool [14]. His Anthropic announcement echoed that framing, describing the current LLM moment as 'especially formative' [15], and acknowledged his ongoing passion for education while flagging it would be deprioritized near-term. Eureka Labs is noted to have raised $6.7M in seed funding in 2026 [16], suggesting the company retains some independent existence, though its operational status going forward has not been clarified.

The community reaction framed the move primarily as a competitive signal. Commentators positioned it as a meaningful win for Anthropic and a symbolic loss for OpenAI [17][18][19][20]. Some viral posts described Karpathy as 'OpenAI's 11th co-founder' now working for a rival [18]; others offered ironic inversions ('Karpathy didn't join Anthropic. Anthropic joined Karpathy.' [21]). Sensationalist framing in social media characterized the hire as Anthropic 'stealing' OpenAI's talent as an 'AI war' escalates [22], while more measured commentary noted that OpenAI's own concurrent announcements quickly redirected the news cycle [23][24]. International coverage in Chinese and Japanese has continued to amplify the competitive framing [25][26]. One fringe claim asserted Karpathy joined specifically to lead 'recursive self-improvement' research [27]; this has not been corroborated by any credible outlet. A separate, context-free social post referenced Karpathy asking for access to something called 'Mythos' at Anthropic [28], which may reflect early engagement with internal infrastructure but lacks sufficient context to interpret.

The hire arrived in the context of Anthropic's reported $30B fundraise at a $900B valuation [29][30] and a claimed first profitable quarter [31], lending it additional financial weight. As of late May 2026, Karpathy's announcement has reportedly accumulated approximately 3 million views [32], and sustained amplification — including retweets of a clip described as 'Karpathy just explained the future of software for free' [33][34][35] — continues to circulate, though without introducing substantive new reporting.

Timeline

  • 2024-02: Karpathy departs OpenAI for the second time [11]
  • 2024-07: Karpathy announces Eureka Labs, an AI-native education company [12][13]
  • 2026-04: Karpathy presents at Sequoia AI Ascent 2026 fireside chat, arguing LLMs are category-creating technologies, not just productivity tools; introduces 'menugen' as a new AI application horizon [14][37][46][47][48]
  • 2026-04: Karpathy publicly signals he might join a major AI lab, citing the desire to be at the frontier of capability [36]
  • 2026-05-19: Karpathy officially announces he has joined Anthropic via Twitter [1][15]
  • 2026-05-19: TechCrunch, CNBC, Axios, WSJ, Forbes, and Reuters confirm Karpathy joins Anthropic's pre-training team specifically [2][3][4][5][6][7]
  • 2026-05-19: AI community reacts; hire framed as competitive win for Anthropic over OpenAI; Rohan Paul surfaces Karpathy's prior foreshadowing comments [17][18][44][36]
  • 2026-05-19–24: Sustained international and social media amplification continues; announcement reportedly reaches ~3M views; reactive coverage in Chinese, Japanese, and English [32][25][26][33][34][35][39][40][21][42][43]
  • 2026: Eureka Labs records $6.7M seed funding round [16]

Perspectives

Andrej Karpathy

Enthusiastic about joining Anthropic; frames the decision around the historical significance of the current LLM moment; signals intent to eventually return to AI education but deprioritizing it near-term; pre-hire public framing casts LLMs as category-creating, not merely productivity-enhancing

Evolution: The Sequoia Ascent content (item 2683) adds intellectual texture to a stance that was previously known only in outline — his bullish, expansive view of LLMs as opening new application categories directly preceded and contextualizes the Anthropic hire

[15][1][36][14][37]

Major tech press (TechCrunch, CNBC, Axios, WSJ, Forbes, Reuters)

Confirm the pre-training team assignment; treat the hire as a significant competitive talent signal favoring Anthropic; Reuters adds to the confirmation chorus

Evolution: Reuters coverage (item 15627) adds another tier-1 wire service to the confirmation list; framing otherwise consistent with prior pass

[2][4][3][5][6][7]

AI community commentators (social media and newsletters)

Broadly frame the hire as a loss for OpenAI and win for Anthropic; sensationalist posts invoke 'AI war' framing; ironic counter-takes circulate alongside ('Anthropic joined Karpathy'); continued amplification of Sequoia talk clips as 'the future of software explained for free'

Evolution: Amplification volume continues; ironic inversions (item 15623) add a new minor register; retweet clusters around the Sequoia clip suggest his pre-Anthropic intellectual framing is now being re-read through the lens of the hire

[17][18][23][27][20][38][22][33][34][35][39][40][32][41][25][21][42][43][26]

Rohan Paul (@rohanpaul_ai)

Congratulatory and contextualizing; surfaces Karpathy's prior public comments as corroborating foreshadowing

Evolution: Consistent; amplifying role

[44][45]

Tensions

  • Confirmed pre-training team role (per TechCrunch, CNBC, Reuters) vs. unverified social media claim that Karpathy will lead 'recursive self-improvement' research — the latter has not been corroborated by any credible outlet [2][10][7][27]
  • Competitive framing ('Anthropic steals OpenAI's best talent') vs. ironic counter-commentary noting OpenAI's own concurrent announcements quickly shifted the narrative — reflecting ongoing instability in who is 'winning' the frontier AI race on any given day [17][23][24][20][22][21]

Sources

  1. [1] I’ve joined Anthropic — reactive:karpathy-joins-anthropic (2026-05-19)
  2. [2] OpenAI co-founder Andrej Karpathy joins Anthropic's pre-training team — reactive:karpathy-joins-anthropic
  3. [3] OpenAI co-founder Andrej Karpathy joins Anthropic - Axios — reactive:karpathy-joins-anthropic
  4. [4] Anthropic hires OpenAI cofounder Andrej Karpathy — reactive:karpathy-joins-anthropic (2026-05-20)
  5. [5] Andrej Karpathy, Tesla Alum and OpenAI Co-Founder, Joins Anthropic — reactive:karpathy-joins-anthropic
  6. [6] OpenAI Cofounder Andrej Karpathy Joins Rival Anthropic - Forbes — reactive:karpathy-joins-anthropic
  7. [7] OpenAI co-founder and former Tesla AI executive Karpathy joins ... — reactive:karpathy-joins-anthropic
  8. [8] Andrej Karpathy joins Anthropic to accelerate Claude pre-training — reactive:karpathy-joins-anthropic (2026-05-20)
  9. [9] Andrej Karpathy Joined Anthropic's Pre-Training Team — reactive:karpathy-joins-anthropic (2026-05-20)
  10. [10] Anthropic hires OpenAI co-founder Andrej Karpathy to lead Claude ... — reactive:karpathy-joins-anthropic
  11. [11] Hi everyone yes, I left OpenAI yesterday - Hacker News — reactive:karpathy-joins-anthropic
  12. [12] Andrej Karpathy is starting Eureka Labs, an "AI Native School" — reactive:karpathy-joins-anthropic
  13. [13] Excited to share that I am starting an AI+Education company called ... — reactive:karpathy-joins-anthropic
  14. [14] Fireside chat at Sequoia Ascent 2026 from a ~week ago. Some highlights: — Andrej Karpathy Twitter (2026-04-30)
  15. [15] Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative.… — Andrej Karpathy Twitter (2026-05-19)
  16. [16] Eureka Labs $6.7M Seed Funding (2026) | Investors & Contacts — reactive:karpathy-joins-anthropic
  17. [17] 🔥 Andrej Karpathy Just Joined Anthropic. Here's Why That Terrifies OpenAI — https://t.co/hLHeNklyI4 — reactive:karpathy-joins-anthropic (2026-05-22)
  18. [18] RT @Dovydas44444: OpenAI lost its 11th co-founder. Anthropic gained him. Andrej Karpathy just joined Anthropic this week... — reactive:karpathy-joins-anthropic (2026-05-22)
  19. [19] Anthropic Lands AI Talent Coup: Andrej Karpathy Joins from OpenAI ... — reactive:karpathy-joins-anthropic
  20. [20] The AI Talent War Is Moving Deeper - Exploring ChatGPT — reactive:karpathy-joins-anthropic
  21. [21] @seelffff Karpathy didn’t join Anthropic. Anthropic joined Karpathy. — reactive:karpathy-joins-anthropic (2026-05-22)
  22. [22] "26 Million Jobs GONE!" - Anthropic STEALS OpenAI's Best As AI ... — reactive:karpathy-joins-anthropic
  23. [23] RT @Andr3jH: "Yesterday Karpathy joined Anthropic so it was over for OpenAI and Anthropic was back but today OpenAI solv... — reactive:karpathy-joins-anthropic (2026-05-22)
  24. [24] RT @Andr3jH: "Yesterday Karpathy joined Anthropic so it was over for OpenAI and Anthropic was back but today OpenAI solv... — reactive:karpathy-joins-anthropic (2026-05-22)
  25. [25] # Andrej Karpathy离开OpenAI加入Anthropic：AI人才战打响了 — reactive:karpathy-joins-anthropic (2026-05-23)
  26. [26] 8. Karpathy の Anthropic 加入を「分水嶺」と読む — 中文圏の深掘り解説 — reactive:karpathy-joins-anthropic (2026-05-21)
  27. [27] 🚀 [Karpathy joins Anthropic to lead recursive self-improvement] Ex-OpenAI co-founder and Tesla AI chief will build RSI r... — reactive:karpathy-joins-anthropic (2026-05-21)
  28. [28] Karpathy: Can I get access to Mythos? — reactive:karpathy-joins-anthropic (2026-05-23)
  29. [29] Andrej Karpathy Joins Anthropic & Anthropic Raises $30BN at $900 ... — reactive:karpathy-joins-anthropic
  30. [30] Anthropic's $900B Valuation Reshapes AI... | Metaintro — reactive:karpathy-joins-anthropic
  31. [31] OpenAI IPO filing day. Trump AI EO signing. Anthropic first profitable quarter. Karpathy at Anthropic. DHARA article ton... — reactive:karpathy-joins-anthropic (2026-05-22)
  32. [32] 👨‍💻 [Karpathy joins Anthropic] OpenAI co-founder announces: "I've joined Anthropic." Joins pre-training team. 3M views i... — reactive:karpathy-joins-anthropic (2026-05-23)
  33. [33] RT @RoundtableSpace: KARPATHY JUST EXPLAINED THE FUTURE OF SOFTWARE FOR FREE — reactive:karpathy-joins-anthropic (2026-05-24)
  34. [34] RT @RoundtableSpace: KARPATHY JUST EXPLAINED THE FUTURE OF SOFTWARE FOR FREE — reactive:karpathy-joins-anthropic (2026-05-23)
  35. [35] RT @RoundtableSpace: KARPATHY JUST EXPLAINED THE FUTURE OF SOFTWARE FOR FREE — reactive:karpathy-joins-anthropic (2026-05-23)
  36. [36] Andrej Karpathy had already talked about this possibility 1 month back, that he might join one of the big AI labs and wh… — Rohan Paul Twitter (2026-05-19)
  37. [37] Sequoia AI Ascent 2026: Andrej Karpathy - by Guillermo Flor — reactive:agentic-coding-debate
  38. [38] Ex-OpenAI Co-Founder Just Joined Anthropic...This is HUGE. — reactive:karpathy-joins-anthropic
  39. [39] RT @RoundtableSpace: KARPATHY JUST EXPLAINED THE FUTURE OF SOFTWARE FOR FREE — reactive:karpathy-joins-anthropic (2026-05-23)
  40. [40] RT @RoundtableSpace: KARPATHY JUST EXPLAINED THE FUTURE OF SOFTWARE FOR FREE — reactive:karpathy-joins-anthropic (2026-05-23)
  41. [41] NEW: Karpathy Joins Anthropic Pre-Training Team — reactive:karpathy-joins-anthropic (2026-05-23)
  42. [42] The Translation Layer How Anthropic's hire of Andrej Karpathy reveals the hidden competition in AI https://t.co/IBmH6wpx... — reactive:karpathy-joins-anthropic (2026-05-22)
  43. [43] The Translation Layer How Anthropic's hire of Andrej Karpathy reveals the hidden competition in AI https://t.co/IBmH6wpx... — reactive:karpathy-joins-anthropic (2026-05-22)
  44. [44] @karpathy Congrats to Anthropic. 🚀 — Rohan Paul Twitter (2026-05-19)
  45. [45] RT @rohanpaul_ai: Andrej Karpathy had already talked about this possibility 1 month back, that he might join one of the … — Rohan Paul Twitter (2026-05-20)
  46. [46] Andrej Karpathy and I are back at Sequoia Capital's AI Ascent 2026 ... — reactive:karpathy-joins-anthropic
  47. [47] AI Ascent 2026 | Sequoia Capital — reactive:karpathy-joins-anthropic
  48. [48] AI Ascent 2026 - YouTube — reactive:karpathy-joins-anthropic