Herrington Darkholme

Announcing the book "𝑀𝑎𝑠𝑡𝑒𝑟𝑖𝑛𝑔 𝑎𝑠𝑡-𝑔𝑟𝑒𝑝"!🦉 The book is a structured deep-dive: we start with the mental models of ASTs, move to rule composition, and end with advanced usage. It’s the place where design principles and philosophy are explained. Link below ↓

As two of the largest forces in equity markets -- growing index ownership and increasing amounts of capital controlled by extremely short-term-oriented, leveraged, volatility-intolerant investors -- converge, we have found occasional opportunities to acquire some of the most dominant long-term compounding franchises at attractive valuations. For example, we acquired Alphabet $GOOG when the stock declined substantially on the release of ChatGPT in late 2022, Amazon $AMZN in the weeks following Liberation Day, and $META more recently on the market's response to the company's unexpectedly large cap ex guidance and expenditures. In our 13F which we will file later today, we will disclose a new position in Microsoft, a company we have followed for many years now offered at a highly compelling valuation. While $PSUS will not be filing a 13F tomorrow, it has also recently made $MFST a core holding. Microsoft operates two of the most valuable franchises in enterprise technology, which account for approximately 70% of the company's overall profits: M365 and Azure. M365, the company's productivity suite, is the dominant operating platform for knowledge work, with over 450 million workers using Word, Excel, PowerPoint, Outlook, and Teams on a daily basis. Azure is the world's second-largest hyperscaler cloud platform and, like AWS in our Amazon investment, is a direct beneficiary of the multi-decade migration of enterprise IT workloads to the cloud, which is now further accelerated by surging demand for AI inference workloads. Both M365 and Azure are underpinned by Microsoft's unparalleled enterprise distribution and the security, compliance, and identity infrastructure it has built and refined over decades. Beyond these core franchises, Microsoft also owns a portfolio of other leading businesses, including LinkedIn (the world's largest professional network with 1.3 billion members), its gaming platform (Xbox and Activision Blizzard), and search and news advertising (Bing and the Edge browser). We began building our position in MSFT in February following a meaningful share price decline after the company reported its fiscal Q2 2026 results. We were able to establish our position at a valuation of 21 times forward earnings, broadly in line with the market multiple and well below Microsoft's trading average over the last few years. Notably, MSFT's headline multiple does not reflect the value of Microsoft's approximately 27% economic interest in OpenAI, which would represent approximately $200 billion, or 7% of Microsoft's market capitalization, at OpenAI's most recent funding round valuation. We believe Microsoft's recent share price decline has been principally driven by investor concerns around two key issues: i) the competitive positioning of M365 against increasingly capable AI lab offerings (notably Anthropic's Claude Cowork), and ii) the durability of Azure's growth, especially in light of Microsoft's evolving relationship with OpenAI. In our view, investors underestimate the resilience of the M365 franchise given its deeply embedded role across enterprises and highly attractive price-value proposition. Unlike point software solutions, which may be vulnerable to disintermediation by better-performing AI alternatives, M365 is tightly integrated into the daily workflow of nearly every large enterprise and is supported by Microsoft's identity, security, compliance, and data governance infrastructure, which would be nearly impossible to replicate. Attractive bundle economics further reinforce Microsoft's advantage, with monthly average revenue per user on the M365 suite at approximately $20, less than half of what customers would pay to purchase the underlying applications individually from different vendors. Moreover, we are encouraged to see Microsoft prioritizing its R&D efforts and investment in Copilot, its own AI agent embedded across M365, with direct involvement from CEO Satya Nadella. We believe these efforts will translate into improved product velocity and greater customer adoption over time. Alongside Copilot's rollout, the company has also begun shifting its pricing model from pure per-seat licensing to a hybrid model of seats plus metered consumption, which helps expand the company’s revenue opportunity as AI agents drive incremental usage that a seat-only structure would not capture. These initiatives should help sustain M365’s strong underlying growth momentum, which was already evident in the business unit’s 15% revenue growth (in constant currency) last quarter. We believe concerns regarding Azure's growth trajectory are similarly misplaced, particularly in light of the franchise's exceptional recent performance. Azure revenue grew 39% in constant currency last quarter, with company guiding to modest acceleration through the second half of the year. We view Microsoft's recent decision to restructure its OpenAI partnership not as a concession but as part of a deliberate pivot toward a more open, multi-model architecture that better serves enterprise customers, who increasingly seek optionality across model providers. Microsoft recently disclosed that over 10,000 enterprise customers have used more than one model on Azure Foundry, the company’s modular AI model marketplace. This model-agnostic approach also strengthens Copilot, which can auto-route queries across multiple models to deliver the optimal output for a given task. To support Azure's rapid growth amid persistent supply constraints, Microsoft has raised its calendar year 2026 capex budget to approximately $190 billion. Consistent with what we have observed at hyperscaler peers Amazon and Google, we view this spend as growth capex that should drive future revenue generation. This is particularly true for Microsoft, given that roughly two-thirds of its capex budget is allocated to server and networking equipment that correlates directly with near-term revenue. Like our purchases of $GOOG, $AMZN, and $META, we believe that $MSFT offers analogous and compelling long-term value at today's valuation.

cute feature

Jay Powell faced a pandemic that nearly broke markets, the worst inflation in 40 years, and a sustained presidential assault on Fed independence. Few Fed chairs have faced one of those. He faced all three. A reflection on what he leaves behind: wsj.com/economy/centra…

@jc_he_afk 学习一下 Bridge 的设计！

让一百个人爱上你的产品，好过让一百万个人“有点喜欢”你的产品。 创业以后，特别认可一句话叫“优秀是卓越的阻碍”。就是说，一个很“优秀”的人，一个很“优秀”的产品很难有动力去把事情做到最后一公里。真正做出好东西的founder，很少有履历一路坦途的985高学历大厂高管。 他们会因为“优秀”的退路太多走向平庸。 在@paulg 那里，这也是老生常谈的所谓“Do things that don't scale”. 做东西不能一开始就求快。 所以，我们思前想后，还是给bridge上了邀请码。因为我们真的想与第一波用户产生深度的联系，和早期的用户学习一下，有哪些事情是“只有bridge能做好”的。 Vibe working tool. 需要在早期有个好点的氛围。 本条推前一百个留言的人，我会送自带管饱token的 bridge 激活码

当你觉得和女朋友无法沟通，不妨看看：

Bun の Rust 移植もう main にマージされててスピード感がすごい． > 2,188 files changed +1,009,257 -4,024 github.com/oven-sh/bun/co…

Fuck $MSFT and everything they have ever invented and all of their products.

Continual learning doesn’t need to tweak neural weights — it can evolve programmatic heuristic rules. Heuristic Learning turns experience into code updates, letting agents grow, refactor, and retain skills like real software systems. trinkle23897.github.io/learning-beyon…

codex is the best AI coding product and we want to make it easy to try. for the next 30 days, we are giving companies that want to try switching over two months of free codex usage.

For every person who replies with a screenshot of their cancelled Claude Code plan, I will donate $10 to open source.

I can't help but feel personally burned by the Claude Code changes announced today. We put so much work into wrapping the (atrocious) Claude Agent SDK in T3 Code. It was the ONLY path they supported, so we made it work. It was hell. Now our users are getting their rate limits cut by 40x, despite us doing everything right. I listened to the Claude Code team. I had my issues with their direction, but I trusted them and took them at their word. I will never make that mistake again. Until we see significant change, it is safe to assume any statement from an Anthropic employee is a lie on a timer. The rug will be pulled, no matter how many promises are made beforehand.

@Mr_Derivatives fixed Tim Apple

AI is getting out of hand lol.

@_4ever00 太羡慕了

Team Building 真的是 E 人天堂，I 人地狱。铃被榨干了 😭

@__oQuery @localhost_4173 事到如今 最对不起的 就是家人

@localhost_4173 MSFT

我是怎么做到大牛市里面，我每天都是红的？

@__oQuery 我是怎么做到大牛市里面，我每天都是红的？

@yusukebe うまそう

朝から！春キャベツのペペロンチーノ

@flrande 大非农数据都已经写明了 11% IT job 没了 layoffs.fyi

天真

SECURITY ADVISORY — TanStack npm packages A supply-chain compromise affecting 42 @tanstack/* packages (84 versions total) was published to npm earlier today at approximately 19:20 and 19:26 UTC. Two malicious versions per package. Status: ACTIVE — packages are deprecated, npm security engaged, publish path being shut down. Severity: HIGH — payload exfiltrates AWS, GCP, Kubernetes, and Vault credentials, GitHub tokens, .npmrc contents, and SSH keys. If you installed any @tanstack/* package between 19:20 and 19:30 UTC today, treat the host as potentially compromised: • Rotate cloud, GitHub, and SSH credentials immediately • Audit cloud audit logs for the last several hours • Pin to a prior known-good version and reinstall from a clean lockfile Detection — the malicious manifest contains: "optionalDependencies": { "@tanstack/setup": "github:tanstack/router#79ac49ee..." } Any version with this entry is compromised. The payload is delivered via a git-resolved optionalDependency whose prepare script runs router_init.js (~2.3 MB, smuggled into each tarball at the package root). Unpublish is blocked by npm policy for most affected packages due to existing third-party dependents. All 84 versions are being deprecated with a SECURITY warning, and npm security has been engaged to pull tarballs at the registry level. Full technical breakdown, complete package and version list, and rolling status updates: github.com/TanStack/route… Credit to the security researcher for responsible disclosure.