Greg Speicher

“the best business to own is one that over an extended period can employ large amounts of incremental capital at very high rates of return.” - Warren Buffett $GOOGL $AMZN

Your money is a commons. It's where you store your life's work — a battery for labor you've already done. And like every commons, it can be grazed. Ecologist Garrett Hardin called it the tragedy of the commons. Shared resources get depleted because whoever takes the most benefits, while everyone shares the cost. Money is no exception: whoever prints it grazes first, and the wage earner gets to the pasture last, after it's been eaten. For fifty years, the dollar has lost purchasing power. The people furthest from the money printer pay for it in dollars that buy less. Elinor Ostrom spent her career looking where Hardin didn't — at the commons that didn't get looted. Swiss pastures, Turkish fisheries, Maine lobster grounds — kept intact for generations by the users themselves. Her finding: a commons survives when the users have a say in the rules, can monitor compliance, and can defend the resource against outsiders who would take control. Nobel, 2009. Fiat money fails all three. The rules change when it's convenient. You can see the money being created. You can't stop it. The grazing shows up in the price of a house, the cost of groceries, the years of work it takes to get ahead. Bitcoin fences the pasture. Fixed supply. Every node checks every block. New coins follow a schedule anyone can verify. No one gets to print more because the rules have become inconvenient. No one can rewrite the rules without you. Ostrom showed a commons can be protected from the people who'd loot it. Bitcoin takes that principle into money itself. The people who run it can enforce the rules. The people who want to change them have to persuade the people who use it. The saver has a veto. #Bitcoin

one og giving props to another og. nice

36th Annual Graham & Dodd Event with Li Lu '96, Founder and Chairman of ... youtu.be/YRhnX_JAbC8?si=0VKC… via @YouTube

Analysts project that this merchant TPU business could generate between $100 billion and $252.7 billion in annual revenue by 2028—a scale that, at the high end, would match the current run rate of Google’s core Search business in just two and a half years. $GOOGL

"So, if the company is better positioned today than it was a year ago, why is the stock price so much lower than it was a year ago? As the famed investor Benjamin Graham said, 'In the short term, the market is a voting machine; in the long term, it's a weighing machine.' Clearly there was a lot of voting going on in the boom year of 1999—and much less weighing. We're a company that wants to be weighed, and over time, we will be—over the long term, all companies are. In the meantime, we have our heads down working to build a heavier and heavier company." This passage, from Bezos's 2000 Amazon shareholder letter, captures where your head should be on a day when everything's red. If you own $AMZN or $GOOGL, close the price chart. Open the KPIs across every business they run — the cloud units and their backlogs, the custom silicon, the ad businesses, the subscriptions, the logistics network, the unit economics. Those charts are up and to the right. That's what you should be looking at. The market is voting today. It will weigh later. Your job is to own businesses that keep getting heavier — and then wait patiently.

Thought provoking essay. (h/t @tylercowen) Four Direct, Reasonable, and Important AI Questions (Which the AI People Will Ignore) freddiedeboer.substack.com/p…

Impressive. What's next? $GOOGL

Google actually doing it. TPUv7 with vLLM and FP8 just dropped a 1.67x perf/$ TCO win over GB300 NVL72 FP4 on AgentX agentic workloads. The Nvidia inference monopoly is officially getting squeezed. Stack control + externalization is real, though CoWoS and power megawatts remain the ultimate gating items. Would love @firstadopter's take. $GOOGL

Impressive demo. Gemini at Work 2026 youtube.com/live/1ASzWklab2U… via @YouTube

Google's Nick Fox on the Future of AI, Jobs and EU Regulation | Popular ... youtu.be/5ufhbisySSU?si=mBjq… via @YouTube

2026 Reunion - The Meeting: Spirit of Teamwork youtu.be/FtT5APeyceM?si=14QF… via @YouTube

Ask Muse to make a reservation for Saturday night. It ground away for several minutes, came back, and told me that the place only takes reservations via telephone. I already knew that, but just wanted to see how Muse would handle it. Am I hallucinating, or were the early reports that it could make phone calls kind of underwhelming, quite frankly?

$MA, five years: EPS compounded 22% a year. The P/E went from 57x to 30x. Shareholders got 7.5% a year. The business did its job. The multiple took the rest. The thesis: Mastercard owns half the world's card network outside China and takes a small toll on $11 trillion of spending. Services like fraud, data and security now grow twice as fast as the core network. Stablecoins and AI agents still need identity, fraud protection and universal acceptance, and Mastercard already supplies all three. Next five years from $551: Base: 14% EPS growth, 25x exit → ~$955, ~12% a year Bull: 17.5% EPS growth, 30x exit → ~$1,340, ~20% a year No multiple expansion required. The market is pricing disruption. The numbers show compounding.

Koray Kavukcuoglu—Google DeepMind SVP and Chief AI Architect, the man now leading Gemini—was just asked on the Google For Developers pod whether Google is committed to being at the frontier of AI. There's been real skepticism on this, including from SemiAnalysis. His answer, on the record: "There's nothing other than being at the frontier that is important for us." He went further. Every decision, every prioritization, every goal runs through that filter. He said he's 100% certain they'll get there—pointing to the team, the full stack, and a company that was building AI chips fifteen years ago, before anyone was talking about AI. And the tell: Gemini 4 is now the most ambitious pre-training run Google has ever done. youtu.be/Rrr2gdbvNFU?si=zcvL… $GOOGL

Jensen's five-layer cake is a useful framework, but it's deficient. It should include, at the base layer, a sixth component: capital — a company's ability to access the capital markets, or better, to self-fund. $NVDA $GOOGL

OpenAI "The company’s operating loss, which includes stock-based compensation, widened from $9.3 billion in the first quarter to $12.3 billion in the second quarter, outpacing its revenue growth." OpenAI’s revenue grew much slower than Anthropic’s last quarter, disappointing investors who had hoped the ChatGPT maker would start to catch up to its rival wsj.com/tech/ai/openais-seco… via @WSJ

OpenAI 2026 departures per Grok... Denise Dresser: Chief Revenue Officer Brad Lightcap: Former COO and longtime executive (since 2018) Fidji Simo: CEO of Applications (de facto No. 2) Joshua Achiam: Chief Futurist Johannes Heidecke: Head of Safety Systems Chloé Bakalar: Head of Ethics Kevin Weil: Former Chief Product Officer / OpenAI for Science lead Bill Peebles: Head of the Sora team Srinivas Narayanan: CTO of B2B applications Kate Rouch: Chief Marketing Officer Barret Zoph: Enterprise AI sales / research leader Caitlin Kalinowski: Robotics and consumer hardware lead Scott Gray: GPU systems expert Where's @firstadopter?

$GOOGL What Is the Floor? Can I have roughly 90% confidence that even the lower end of Alphabet's plausible future produces an adequate investment return? And in the genuinely bad 10% tail, is the company not merely alive but still economically valuable to its owners? This should remain the first question because everything else is subordinate to it.

Been reading through some @GrahamDuncanNYC essays. They are packed with wisdom. grahamduncan.blog/the-infini…