# Zuckerberg's Superintelligence Letter Draws Five Founder Endorsements

*By Recommended Reading from Tech Founders • July 29, 2026*

Mark Zuckerberg's piece on a positive vision for superintelligence, emphasizing broad access over concentrated control, was endorsed by five tech leaders—making it the period's strongest multi-recommender signal. Plus formative reads from Palmer Luckey, Sarah Guo, Garry Tan, David Perell, and Amjad Masad.

## Most compelling: Zuckerberg's superintelligence letter

- **Content type:** Article / letter (shared on X)
- **Author:** Mark Zuckerberg
- **Link:** [x.com/finkd/status/2082160210399948869](https://x.com/finkd/status/2082160210399948869)
- **Recommended by:** Aaron Levie, Tobi Lütke, Alexandr Wang, Satya Nadella, David Sacks

Zuckerberg wrote that "we believe the future is for everyone," previewing a positive vision for a world with superintelligence [^1]. Five prominent tech leaders independently endorsed the piece within hours—the period's strongest multi-recommender signal.

Alexandr Wang distilled its guiding principles as "individual empowerment, invention over automation, and balance of power through broad access rather than concentrated control" [^2]. David Sacks quoted Zuckerberg directly: "The defining question of our age isn't whether superintelligence will exist, but who will have access to it. Will it be centralized and restricted to a few institutions, or will it be a tool that empowers everyone?" Sacks extended the argument, calling concentration of power "the biggest risk of AI" and framing broad access and transparency as the path to actual security rather than regulatory capture [^3].

Aaron Levie called it a "Great piece and vision for AI" [^4]. Tobi Lütke called it a "great read" [^5]. Satya Nadella tied it to building "a frontier ecosystem that empowers people and orgs everywhere" [^6].

**Why it matters:** Five leaders across competing AI ecosystems—Meta, Microsoft, Shopify, Box, and Scale AI—converged on the same decentralization argument in the same window. The endorsement pattern itself is the signal: the open-access-versus-control framing is becoming a shared industry position, not a single company's talking point.

## A technical walkthrough: "22580: From GPT2 to Kimi3, Explained"

- **Content type:** Technical article / worklog
- **Author:** @waterloo_intern (described by Guo as baseten's "intern")
- **Link:** [x.com/i/article/2077616768491585536](https://x.com/i/article/2077616768491585536)
- **Recommended by:** Sarah Guo

Sarah Guo called it a "Great write up on technical innovations in Kimi3 that all can use" [^7]. The article traces major architectural developments from GPT-2 (2019) to KimiK3 (2026), framing the journey around a single number: 22,580 GPT-2 models fit inside one KimiK3, a scale-up factor achieved in seven years. It then asks whether that progress is "just... scale" or reflects genuine architectural change, walking through decoder-only architecture, KV caching, and memory mechanisms along the way [^8].

**Why it matters:** Guo's endorsement emphasizes practicality—"innovations that all can use"—making this a directly accessible technical resource for builders tracking how frontier model architecture has evolved, rather than a marketing announcement.

## Palmer Luckey's formative read: "Assassination Politics" by Jim Bell

- **Content type:** Essay (1996)
- **Author:** Jim Bell
- **Originally published on:** Outpost of Freedom website
- **Recommended by:** Palmer Luckey (in a YouTube interview)

In a long-form interview, Luckey described reading Jim Bell's 1996 essay "Assassination Politics" as a pivotal moment. Bell predicted the emergence of "a cryptographically verifiable semi-anonymous cryptocurrency" and argued it would make large-scale oppressive government unfeasible through anonymous crowdsourced bounty systems [^9]. Luckey said the essay prompted him to search for whether such a technology actually existed—leading him to Bitcoin at its ground floor, before any exchanges existed [^9].

**Why it matters:** Luckey credits a specific essay as the intellectual bridge between a science-fiction idea and his earliest Bitcoin involvement. It is a rare first-person account of the reading that triggered a consequential early bet.

## Garry Tan's pick: *Nothing Left* by Evan W. Chen

- **Content type:** Book (political memoir)
- **Author:** Evan W. Chen
- **Link:** [x.com/evanwch/status/2082093236063998261](https://x.com/evanwch/status/2082093236063998261)
- **Recommended by:** Garry Tan

Tan endorsed the book as "Important" for "explaining what is going wrong with the Dem party and how we bring it back to common sense" [^10]. Chen described it as his political memoir, released the same day [^11].

**Why it matters:** Tan's recommendation is concise but deliberate—a Y Combinator leader singling out a political memoir on launch day signals he sees it as relevant to the current policy moment, not a casual mention.

## David Perell's pick: the Michael Ovitz biography

- **Content type:** Book (biography)
- **Author:** Not specified by Perell
- **Recommended by:** David Perell

Perell said the most striking part of the Ovitz biography was learning the upper limits of corporate gift giving: Ovitz's gifts office spent roughly $500,000 per year, one of five assistants was solely responsible for gifts, and the guiding rule was "no disposable gifts"—nothing rinse-and-repeat like wine or champagne, but bespoke items such as first edition books or collectible coins tailored to each recipient's interests [^12].

**Why it matters:** Perell extracted a specific, actionable principle—personalized over generic giving—from a biography, which is a stronger endorsement signal than a passing title mention. The detail about a dedicated gifts assistant tracking clients' hobbies and charities makes the practice concrete.

## Amjad Masad's pick: the Erdős problems thread

- **Content type:** X thread
- **Author:** @Qiaoqiao2001
- **Link:** [x.com/Qiaoqiao2001/status/2080003441821163958](https://x.com/Qiaoqiao2001/status/2080003441821163958)
- **Recommended by:** Amjad Masad

Masad recommended a thread in which @Qiaoqiao2001 described solving six open Erdős problems in five days using OpenAI's GPT-5.6 Sol, with a Codex workflow that "does not require deep mathematical knowledge" [^13]. Masad framed it as evidence of a new era: "Our generation may explore the computational universe: the vast space of algorithms, programs, proofs, and designs that AI agents can search" [^14].

**Why it matters:** Masad's endorsement elevates a practitioner thread into a thesis about AI as an exploration tool for mathematics. The thread includes the author's prompts and workflow, making it a reproducible resource rather than a claim.

---

All items above are organic recommendations made in social posts or long-form interviews; no sponsored, paid, or self-promotional material is included.

---

### Sources

[^1]: [𝕏 post by @finkd](https://x.com/finkd/status/2082160210399948869)
[^2]: [𝕏 post by @alexandr_wang](https://x.com/alexandr_wang/status/2082170165781778901)
[^3]: [𝕏 post by @DavidSacks](https://x.com/DavidSacks/status/2082226325621035131)
[^4]: [𝕏 post by @levie](https://x.com/levie/status/2082168124733116537)
[^5]: [𝕏 post by @tobi](https://x.com/tobi/status/2082203290809377259)
[^6]: [𝕏 post by @satyanadella](https://x.com/satyanadella/status/2082177964675133931)
[^7]: [𝕏 post by @saranormous](https://x.com/saranormous/status/2082089083120041997)
[^8]: [𝕏 article by @waterloo_intern](https://x.com/i/article/2077616768491585536)
[^9]: [Palmer Luckey on What He Collects, Mining Bitcoin and Building Oculus](https://www.youtube.com/watch?v=zix7qUDrnv8)
[^10]: [𝕏 post by @garrytan](https://x.com/garrytan/status/2082153334115590173)
[^11]: [𝕏 post by @evanwch](https://x.com/evanwch/status/2082093236063998261)
[^12]: [𝕏 post by @david_perell](https://x.com/david_perell/status/2082117317497086324)
[^13]: [𝕏 post by @Qiaoqiao2001](https://x.com/Qiaoqiao2001/status/2080003441821163958)
[^14]: [𝕏 post by @amasad](https://x.com/amasad/status/2082000490066592127)