ARK is hiring a 10x engineer to help scale the capabilitites of our reserach team. If you're an LLM and data ninja, apply here and drop me a DM with the coolest AI/data project you've worked on. app.trinethire.com/companies/6974…

OpenAI's o3 and o4-mini release underscored the importance of productization and tool use to success in the LLM space. Read more about OpenAI's strategy, AI in oncology, Tesla's earnings, and ARK's bitcoin price target in this week's newsletter:

Is the worst behind us? ARK is optimistic that the Fed-induced rolling recession is giving way to an innovation-led recovery. Read a letter from CEO/CIO Cathie Wood to learn why.   ark-invest.com/articles/marke…

I'm very excited to see what impact Meta's new API may have on open-weight model usage and customization, especially given their inference offerings from Cerebras and Groq Inc. More on what these collaborations mean in this week's newsletter from Varshika Prasanna and me:

In the 6 months between the general release of OpenAI's o1 model and the price cuts to o3 last week, the cost of inferencing reasoning models has fallen at a 98% annualized rate. Read more about what we believe the impact of this will be in this week's newsletter:

Really enjoyed this conversation with Dr. Alan Bekker and Brett Winton! If you want to learn more about how AI agents are disrupting the education industry and the way we work more broadly, give it a listen:

Grok 4 shows serious progress on scientific reasoning and tool use (including impressive results on Andon Labs' Vending-Bench). We think tool use and the upcoming coding capabilities xAI is working on will motivate developers to experiment more with Grok for enterprise use:

Visitors to Replit ⠕ seemed to be flatlining last year, but after releasing Replit Agent in September, user growth inflected as people realized how powerful vibe coding can be! Agentic AI enables anyone to become a developer, regardless of experience. (data from Similarweb)

Had a great conversation with Nick Grous and Sam Korus about how AI models like Grok 4 are making progress towards practical adoption. Cost-effective and real-world usefulness will be needed to justify the massive infrastructure buildouts by companies like xAI and Meta .

In the last six months the minimum cost to use models that score at least 50% on Artificial Analysis’s Coding Index has fallen 98.7%, shown below. Read about how intelligence is exploding as costs plummet in Frank Downing's new research blog: ark-invest.com/articles/analy…

Incredible results from Palantir. Alex Karp thinks they could 10x their US revenue in the next 5 years. That sounds shocking, but it's consistent with our belief that leaders in the platform-as-a-service layer will grow at an accelerated rate as AI transforms enterprises:

GPT-5 isn't just an advance in performance- it's a big step forward for usability. Read more about how OpenAI is bringing vibe coding capabilities to their 700M+ weekly users in this week's newsletter:

Perplexity's bid for Chrome underscores how important distribution is for AI model companies, even though their bid may undervalue Chrome by an order of magnitude.

We're super excited to test Replit ⠕'s Agent 3! Increasing the amount of time the agent can run by 10x could be a major capability unlock as agents evolve from a productivity-boosting tool to an autonomous co-worker.