Gianfranco's October 2025 Reading List

The top essays on artificial intelligence, business, technology & society, and self-improvement.

What’s worth reading this month—and why does it matter?

Each day, I devote three hours to reading. Occasionally an author writes a piece that feels like a gift—crafted with intention for the reader. It tends to be a piece that encourages a reader to pause, invites them to feel how a strong idea, delivered with empathy, can resonate.

Few writers achieve this.

Each selection did one of three things for me: it sharpened a thesis, challenged a prior, or opened a new line of inquiry. That’s the bar for inclusion.

In this curation practice, I believe the hero of this story isn’t the author of any single essay, nor me as curator, but you—the reader.

If these pieces don’t move you, I haven’t honored the attention you’ve given me. Hold me to that.

Use this as a tasting menu: no set order, categories for context, and a brief note on what each means for builders and capital allocators.

AI Systems & Interfaces

  • How to Build an AI Data Center - Brian Potter — Construction Physics
    • Potter explains that training large AI models demands physical infrastructure: GPT-4 required an estimated 21 billion petaFLOP operations, and running it on an iPhone would take 60 000 years. Hyperscalers are spending tens of billions to build AI-optimized data centers, but electricity availability has become the binding constraint as a single center can draw as much power as a small city. The essay urges leaders to invest in firm power and domestic build-outs because national competitiveness now depends on marrying digital ambition with industrial capacity.
    • Read More
  • RL isn’t the Silver Bullet for AI-Powered Chip Design - Zach — zach.be
    • Zach argues that large language models struggle at high-performance chip design because Verilog demands precise, performance-sensitive code, and reinforcement learning (RL) doesn’t save the day. RL requires fast simulation environments, but current EDA tools (synthesis, place-and-route, analog simulation) take hours or days. Emerging companies like Silimate, Partcl and Dash Crystal are building AI-accelerated PPA prediction and GPU-powered design tools; once simulation speeds up, RL may help, but until then the real progress comes from improving the tools themselves.
    • Read More

Operator Playbooks

  • An Engineering History of the Manhattan Project - Brian Potter — Construction Physics
    • Brian Potter reframes the Manhattan Project as an industrial-engineering feat rather than a pure science triumph. Wartime urgency and unlimited funding meant pursuing parallel paths — different fuels, separation methods and bomb designs — until a mid-project pivot to an implosion design forced rapid invention in explosives, metallurgy and instrumentation. The leadership lesson is to place multiple costly bets up front and pivot decisively when the facts change.
    • Read More
  • Airbnb’s untapped monetization lever(s) - MBI Deep Dives
    • MBI Deep Dives argues that Airbnb’s biggest opportunity lies in tapping new host services: one-third of bookings already come from rural small-town stays, and a nascent co-host network (currently ~2 % of nights booked) hints at massive growth. Brian Chesky envisions Airbnb offering pricing guidance, cleaning, registration, photography and even an “AI host agent,” while promoted listings could give hosts paid ways to stand out. The essay suggests leaders must test multiple monetization paths and build a real-world social network that deepens community.
    • Read More

Market Structure & Moats

  • Facebook is Dead; Long Live Meta - Ben Thompson — Stratechery
    • Ben Thompson observes that Meta’s blowout earnings coincided with Mark Zuckerberg’s aggressive pivot from the declining Facebook social network to AI-driven products and the metaverse; the company’s advertising juggernaut is funding a new platform strategy. By shifting focus to algorithmic feeds, messaging and hardware, Meta is betting that AI can redefine its relevance even as the original Facebook brand fades. The leadership takeaway is that incumbents must cannibalize their own legacy products to pursue bold new opportunities, accepting near-term uncertainty for long-term dominance.
    • Read More
  • Understanding Figma - Muji — hhhypergrowth
    • Muji chronicles how Figma took three-plus years to build a browser-based design tool whose cloud architecture enabled real-time collaboration—an advantage that let it capture over 80 % share in UI/UX design and 70 % in prototyping. Its product-led, bottom-up motion translated into revenue of $228.2 M (46 % growth) with 91 % gross margins and 13 M monthly users; after the aborted $20 B Adobe acquisition, Figma remains independent and is expanding into a multi-product platform with AI features and enterprise-grade tooling. The story illustrates how betting on new vectors (collaboration and the browser) can invert industry incumbents.
    • Read More
  • Open Banking and Payments Competition - Patrick McKenzie — Bits About Money
    • Patrick McKenzie explains that debates over Section 1033 of the Dodd-Frank Act are less about data sharing and more about payments: banks want to charge fintechs for access to “Open Banking” data because account-to-account transfers threaten lucrative card interchange fees. Open banking lets consumers log into their bank at checkout so fintechs can initiate ACH debits, making payments cheaper and smoother; aggregators like Plaid replaced risky screenscraping with secure API access. Banks are fighting the CFPB’s pro-competition rulemaking through lawsuits and lobbying, while card networks view aggregators as existential threats (Visa tried to buy Plaid to neutralize it), illustrating how entrenched players resist new payment rails that empower consumers and innovators.
    • Read More

Capital & Macro

  • AI Will Not Make You Rich - Join Colossus
    • Join Colossus compares generative AI to containerization, arguing that while transformational technologies like railroads or microprocessors enriched investors, many revolutions primarily benefited customers. In mature tech waves, incumbents capture most gains; today’s ICT giants absorb AI’s value, leaving limited upside for new startups and generalist investors. The takeaway for investors is to look downstream at businesses enabled by AI rather than expecting the AI builders themselves to produce outsized returns.
    • Read More
  • When is Institutional Capital Right for a New VC Firm? - Hunter Walk
    • Hunter Walk advises emerging fund managers to seek institutional LPs only if they’re building a long-term firm, prefer performance risk over perpetual fundraising, and are willing to invest in selecting values-aligned partners. He dispels myths that institutional capital turns LPs into meddlesome customers or erodes flexibility; high-quality LPs add minimal overhead and expect managers to adapt strategy as markets evolve. The leadership lesson is to choose partners deliberately and embrace accountability rather than chasing easy money.
    • Read More
  • Tech Can Fix Most of Our Problems (if we let it) - Noah Smith — Noahpinion
    • Smith argues that technology—AI included—can solve many societal problems if allowed to flourish; studies show GPT-4 chats reduced conspiracy-theory belief by 20%, demonstrating AI’s potential to improve discourse. He notes that COVID-19 vaccines ended the pandemic and renewable technologies are cutting emissions faster than climate politics or activism. The essay urges leaders to champion innovation rather than rely solely on politics, balancing optimism about engineering breakthroughs with awareness of ethical responsibilities.
    • Read More

People & Purpose

  • How to Use AI Without Becoming Stupid - Cedric Chin — Commoncog
    • Cedric Chin introduces the “Vaughn Tan Rule”: never outsource your subjective value judgments to AI without a clear reason, because meaning-making is uniquely human. AI can transcribe, summarise and suggest, but it cannot decide what is worthwhile; even using AI to filter emails trades depth for convenience, so one must explicitly accept the trade-offs. Blaming AI for outcomes is like blaming a hammer; leaders must remain the responsible agent, honing their character and tolerating the discomfort of hard choices.
    • Read More
  • Why I Left Bloomberg - Noah Smith — Noahpinion
    • Noah Smith explains that chronic vestibular migraines limited his screen time, forcing him to choose between his Substack hobby and his day job. He also cites mounting bureaucracy at Bloomberg Opinion, which added layers of editors and decreased joy in writing, and hints at an undisclosed incident involving the Chinese Communist Party. The reflection underscores the trade-offs between health, autonomy and institutional constraints.
    • Read More

Reply

or to participate.