ctaio.dev Ask AI Subscribe free

AI · AI Education · AI accounts on X

Best AI Influencers to Follow on X in 2027

The AI accounts on X most worth a technology leader's time in 2027 are Simon Willison, Andrej Karpathy, Ethan Mollick, Hamel Husain and Dylan Patel, followed by ten more researchers, builders, analysts and lab leaders listed below. I picked them because each one posts original work, such as code, benchmarks, research or cost analysis, rather than reposting the day's news. Follower counts and post totals were read from each X profile page on 19 November 2026, and I list contact routes only where the person publishes one. Follower count did not decide the ranking: the smallest account here, Hamel Husain at 67K, is more useful to most engineering teams than accounts thirty times its size.

By · Published November 19, 2026 · Checked November 19, 2026

A smartphone glowing on a dark desk under amber lamp light, suggesting a curated feed of AI accounts

Key takeaways

  1. Start with five: Simon Willison for practical model behaviour and AI security, Karpathy for mental models, Mollick for adoption research, Hamel Husain for evals and Dylan Patel for infrastructure economics.
  2. Treat lab leaders such as Sam Altman and Demis Hassabis as announcement feeds. Confirm anything you plan around in the lab's release notes or documentation.
  3. Put these accounts in a private X list and read it on a schedule. The algorithmic home feed will bury the posts you followed them for.
  4. Most accounts here have a newsletter or blog behind them. When a thread matters, the long-form version is where the caveats live.
  5. Several excellent people post rarely or have left X. For them, follow the blog or newsletter instead and keep the feed for daily voices.

Which AI accounts on X are worth following in 2027?

Ranked within each group by how much a technology leader gets out of it. "Checked" says where each number came from and when.

Researchers

  1. 4.46M followers · 10K+ posts since 2009

    For: Engineering leaders who want to understand how LLMs work and where hands-on coding with them is heading.

    Founding member of OpenAI and former Director of AI at Tesla, now teaching LLM fundamentals on YouTube. His own site says he spends most of his time on X or GitHub, so the feed is his main channel. Threads tend to turn a vague trend into a mental model your team can reuse.

    Checked: X profile page (follower and post counts), 19 Nov 2026; bio from karpathy.ai

  2. 757K followers · 25K+ posts since 2009

    For: CTOs and CAIOs who need a calibrated counterweight to AGI marketing.

    Creator of Keras and the ARC-AGI benchmark, co-founder of Ndea and the ARC Prize. He argues in public about what current models can and cannot generalise to, with benchmark results behind the argument. Useful when a vendor deck claims reasoning your evals do not show.

    Checked: X profile page (follower and post counts), 19 Nov 2026; email from fchollet.com

  3. 606K followers · 4K+ posts since 2012

    For: Leaders tracking robotics, world models and physical AI.

    NVIDIA Director of Robotics and Distinguished Scientist, co-lead of the GEAR lab. He posts less often than the builders on this list, but each thread usually explains a robotics or foundation-model result in plain terms. The best single feed on X for the physical AI side of the field.

    Checked: X profile page (follower and post counts), 19 Nov 2026; email from jimfan.me

  4. 540K followers · 20K+ posts since 2012

    For: Engineers and architects who want model internals explained with code.

    LLM research engineer, former statistics professor and author of Build a Large Language Model (From Scratch). His posts walk through new architectures and training papers, often linked to long-form write-ups in his Ahead of AI newsletter. The most reliable feed here for knowing what changed inside an open-weight model release.

    Checked: X profile page (follower and post counts), 19 Nov 2026; bio from Ahead of AI About page

Builders

  1. 257K followers · 61K+ posts since 2006

    For: Every engineering leader with LLMs or coding agents in production.

    Co-creator of Django and creator of Datasette and the LLM command-line tool, who coined the terms prompt injection and the lethal trifecta. He writes about LLMs, coding agents and AI security on a blog he has kept since 2002, and his X feed points to that work. If you follow one builder for AI security and practical model behaviour, follow him.

    Checked: X profile page (follower and post counts), 19 Nov 2026; bio from simonwillison.net/about

  2. 215K followers · 79K+ posts since 2009

    For: Heads of platform and AI engineering building out an AI engineer function.

    Runs Latent Space, is affiliated with the AI Engineer conferences, and wrote the essay The Rise of the AI Engineer. At 79K+ posts his feed is high volume, covering tooling, agents and lab releases from an engineer's point of view. Read it through a list and keep the links.

    Checked: X profile page (follower and post counts), 19 Nov 2026; email from latent.space

  3. 67K followers · 18K+ posts since 2012

    For: Anyone who owns quality for an LLM product: eng managers, product leads, QA.

    Machine learning engineer with Airbnb and GitHub behind him, whose X bio now reads "Evals Evals Evals @OpenAI". He co-teaches the AI Evals for Engineers and PMs course, which his site says has had over 5,000 students. The smallest account on this list and one of the most useful.

    Checked: X profile page (follower and post counts), 19 Nov 2026; course figures from hamel.dev

  4. 141K followers · 17K+ posts since 2014

    For: Teams choosing an agent framework or agent observability stack.

    Co-founder of LangChain, which started as his side project in late 2022. He speaks publicly about agent architecture and context engineering for long-running agents, with an obvious LangChain angle you should discount. Worth following for how a framework vendor sees agent patterns changing.

    Checked: X profile page (follower and post counts), 19 Nov 2026; company history from langchain.com/about

  5. 334K followers · 65K+ posts since 2010

    For: Leaders who care about practical AI tooling, open models and AI policy.

    Co-founder of fast.ai and Answer.AI, founding president of Kaggle and creator of the ULMFiT method. With 65K+ posts he is among the most prolific accounts here, covering developer tooling, open models and AI policy. Read him for a practitioner's view from outside the big labs.

    Checked: X profile page (follower and post counts), 19 Nov 2026; bio from jeremy.fast.ai

Analysts and investors

  1. 423K followers · 37K+ posts since 2009

    For: CAIOs, CIOs and anyone explaining AI adoption to a board or an executive team.

    Wharton professor studying AI at work, author of Co-Intelligence and its successor Co-Existence, and writer of the One Useful Thing newsletter. His newsletter offers a research-based view of AI's effect on work and education, and he describes himself as very active on X. The account I'd forward to a CEO first.

    Checked: X profile page (follower and post counts), 19 Nov 2026; bio from oneusefulthing.org/about

  2. 174K followers · 14K+ posts since 2018

    For: Anyone signing off on GPU capacity, neocloud contracts or data center plans.

    Founder of SemiAnalysis, the independent research firm on semiconductors, AI infrastructure and data centers. His posts preview chip, cluster and inference cost analysis that matters for capacity planning. The institutional research is paid, but the feed alone sharpens vendor negotiations.

    Checked: X profile page (follower and post counts), 19 Nov 2026; firm details and email from semianalysis.com

  3. 111K followers · 12K+ posts since 2014

    For: Leaders deciding between closed APIs and open-weight models.

    Executive Director of Trillium Labs, a nonprofit for open frontier AI science, previously post-training lead on Ai2's Olmo models, and author of the Interconnects newsletter and the RLHF Book. He tracks open-model releases, including China's labs, with first-hand training experience. The best feed for judging whether an open model is good enough for your workload.

    Checked: X profile page (follower and post counts), 19 Nov 2026; bio from natolambert.com and interconnects.ai/about

  4. 168K followers · 15K+ posts since 2011

    For: CTOs and corporate development leads watching which AI startups are worth buying from or buying.

    Founder of Conviction, a venture firm that invests in AI-native companies (its portfolio lists Baseten, Cartesia and Cognition), and co-host of the No Priors podcast. Her feed shows where early-stage money is going before it appears in analyst reports. Read it as an investor's view, with the incentives that come with it.

    Checked: X profile page (follower and post counts), 19 Nov 2026; firm details from conviction.com

Lab leaders

  1. 6.45M followers · 8K+ posts since 2006

    For: Anyone whose roadmap depends on OpenAI models or pricing.

    CEO of OpenAI and the most-followed account on this list. Treat it as an announcement channel: product launches, capacity issues and strategic signals often surface here first. Confirm details in OpenAI's release notes before you plan around them.

    Checked: X profile page (follower and post counts), 19 Nov 2026; CEO role per a Nov 2026 news article (blog.mean.ceo, third-party)

  2. 1.97M followers · 3K+ posts since 2013

    For: Leaders on Google Cloud or tracking AI for science.

    Nobel laureate and Google DeepMind co-founder; his X bio lists him as Chair of Google DeepMind, Chief Scientist of Alphabet and CEO of Isomorphic Labs. Low volume, high signal: Gemini milestones and science results such as drug discovery work. Read the linked DeepMind posts rather than the reactions underneath.

    Checked: X profile page (follower and post counts), 19 Nov 2026; roles as stated in his X bio

How did I choose these AI accounts on X?

Every account had to clear three tests. The person posts their own work or first-hand analysis, they post often enough to matter in a weekly reading habit, and their posts change a decision a CTO, CIO or CAIO makes: which model to trust, how to evaluate it, what to pay for compute, how to explain adoption to a board. Commentators who mostly summarise other people's announcements did not make the cut, however large their audience.

I checked every follower count and lifetime post total on the person's X profile page on 19 November 2026, and roles against their own website, newsletter or company page. Post totals are a rough proxy for activity, which is why I give them: an account with 60,000 posts behaves very differently in your feed from one with 3,000. I do not report engagement or posting frequency because I could not verify those reliably.

The groups matter more than the order inside them. Researchers explain what models can do and why. Builders show what breaks when you ship. Analysts and investors cover cost, capacity and capital. Lab leaders tell you what is coming from the vendors you already depend on. A balanced feed needs at least one voice from each group, or you end up with either hype or cynicism and no ground truth.

Which accounts should a CTO, CIO or CAIO follow first?

It depends on the decision on your desk. If you run engineering and your teams are shipping LLM features or rolling out coding agents, start with Simon Willison, Hamel Husain and swyx. Willison covers model behaviour and the security failure modes behind prompt injection, Husain covers the evaluation discipline described in our guide to testing and evaluating AI agents, and swyx covers the tooling and the people defining the AI engineer role.

If you own an AI strategy or a Chief AI Officer mandate, add Ethan Mollick and Nathan Lambert. Mollick turns research on AI at work into arguments an executive team will accept, and Lambert tells you whether an open-weight model is close enough to the frontier to change your build-versus-buy decision. If you sign infrastructure budgets, Dylan Patel is the one account worth reading before every capacity negotiation, alongside our explainer on neoclouds.

For direction of travel, Karpathy, Chollet and Raschka cover the research end from three angles: Karpathy builds intuition, Chollet tests claims against benchmarks, Raschka explains architectures in code. Jim Fan is the specialist pick for anyone with robotics or physical AI on the roadmap. Sarah Guo matters most to leaders who buy from, partner with or acquire early-stage AI companies.

How do you follow AI on X without losing hours a week?

Read a private list instead of the home feed. Add the accounts on this page to an X list, pin it, and read the list once or twice a week. The home feed optimises for engagement, which on AI topics means arguments, screenshots and benchmark fights, while a list shows you what these specific people posted in order.

Click through to the long version. Most people here publish the long version elsewhere: Willison's blog, Mollick's One Useful Thing, Lambert's Interconnects, Raschka's Ahead of AI, swyx's Latent Space. A thread is the headline. The caveats, methods and numbers are in the post, and you should read that before forwarding a claim to your team. Our AI newsletters hub covers that long-form layer.

Discount for incentives. Lab leaders announce their own products, founders post about their own frameworks and investors talk up their portfolios. That does not make them wrong, and several are the earliest source of real news. It does mean that a launch post from a lab CEO is a prompt to read the release notes and run your own evals before a roadmap changes.

What should you do with what you read on X?

Turn posts into tests. When Willison documents a new prompt injection pattern, ask your security lead whether your agents are exposed. When Husain describes an eval failure mode, check whether your own evals would catch it. When Patel publishes inference cost numbers, compare them with your current contract. A feed earns its place when it produces a question for a specific person on your team.

Keep a short running note of claims you plan to verify, with the account and date. After a quarter you will know which accounts were right early and which ones generated noise, and you can prune the list on evidence. Recheck the list every six months, because roles change fast: Nathan Lambert now runs Trillium Labs after leading post-training at Ai2, and Hamel Husain's X bio now lists OpenAI.

Who is missing from this list, and why?

Some obvious names are absent for practical reasons. Benedict Evans, whose technology analysis many executives read, has stated in his X bio that he left the platform; read his newsletter instead. Chip Huyen, author of AI Engineering, has fewer than 600 posts on X in total, and Lilian Weng of Lil'Log has about 220. Both are essential reading, but on their blogs, where they do their real work.

Others post too rarely for a daily feed or overlap with someone already listed. Andrew Ng has close to 2M followers but around 2,000 posts in total, so his letter in the weekly Batch newsletter from DeepLearning.AI is the better channel. Yann LeCun, now founder and chair of AMI Labs, and Logan Kilpatrick, who works on the Gemini API at Google, are both worth adding if world models or Gemini are central to your stack. I kept the core list to fifteen so it fits in one weekly read.

Listed here, or should be?

If your entry is out of date or you think something belongs on this list, email [email protected]. Entries are editorial and free. A featured, labelled Sponsored slot is available through vendor listings.

Related questions

More from AI Education

Frequently Asked Questions

Who is the best AI account to follow on X for a CTO?
For most CTOs it is Simon Willison. He writes about LLMs, coding agents and AI security, created the LLM command-line tool and coined the term prompt injection, so he covers both capability and security. Add Hamel Husain if your team owns an LLM product's quality.
Are follower counts a good way to judge AI influencers on X?
No. Follower counts reflect fame and timing more than usefulness. On this list Hamel Husain has about 67K followers and Sam Altman about 6.45M, yet for an engineering team building LLM features Husain's posts on evals are far more actionable. Judge accounts by whether they publish original work you can check.
How were the follower numbers on this page checked?
Each follower count and lifetime post total was read from the person's public X profile page on 19 November 2026 and rounded. Roles and affiliations were checked against the person's own website, newsletter or company page, or their X bio where noted. Counts change daily, so treat them as a snapshot.
Should I follow AI researchers or AI builders on X?
Both, in proportion to your job. Researchers such as Karpathy, Chollet and Raschka explain why models behave as they do and where the field is heading. Builders such as Willison, Husain and Chase show what breaks in production. Leaders who ship software should weight builders more heavily.
Is X still the best place to follow AI news in 2027?
It is still where many researchers and builders post first, and Karpathy's own site says he spends most of his time on X or GitHub. Some well-known analysts have left the platform, so pair a curated X list with newsletters and blogs rather than relying on the feed alone.
·
Thomas Prommer
Thomas Prommer Technology Executive — CTO/CIO/CTAIO

These salary reports are built on firsthand hiring experience across 20+ years of engineering leadership (adidas, $9B platform, 500+ engineers) and a proprietary network of 200+ executive recruiters and headhunters who share placement data with us directly. As a top-1% expert on institutional investor networks, I've conducted 200+ technical due diligence consultations for PE/VC firms including Blackstone, Bain Capital, and Berenberg — work that requires current, accurate compensation benchmarks across every seniority level. Our team cross-references recruiter data with BLS statistics, job board salary disclosures, and executive compensation surveys to produce ranges you can actually negotiate with.