Picks and shovels win again

Micron Technology reported record Q3 FY26 revenue of $41.16 billion, up 74% sequentially and 346% year-over-year, with net income surging 1,400% to $28.2 billion. The memory maker's strong performance, driven by demand for high-bandwidth memory used in AI compute, alleviated investor fears about a slowdown in data center expansion. Micron also announced 16 strategic customer agreements securing approximately $100 billion in minimum revenue through 2030.

Picks and shovels win again Micron crushes, inflation ticks higher, and your next laptop just got more expensive Welcome to . Cautious Optimism https://www.cautiousoptimism.news/ , a newsletter on tech, business, and power. Modestly upbeat Thursday. Micron reported earnings https://investors.micron.com/news-releases/news-release-details/micron-technology-inc-reports-record-results-third-quarter after the bell yesterday. The company’s third-quarter report was closely watched thanks to how closely Micron is tied to the global compute buildout. If Micron had reported soft earnings and slack guidance, investors may have inferred that future data center capacity expansion would undershoot expectations. If the rush to build more compute capacity was losing momentum, then a host of companies, from foundation model companies to power-generating startups, from young tech companies building agentic orchestration to increasingly levered hyperscalers, may have found the limits of market demand for AI products earlier than anticipated. Micron makes high-bandwidth memory HBM , which helps GPUs avoid idle time as data passes between chips. If you want good compute yield on your GPU spend, you must use HBM other bottlenecks exist, of course . Yes, the Idaho-based memory maker, which exited the consumer market https://investors.micron.com/news-releases/news-release-details/micron-announces-exit-crucial-consumer-business to double down on selling to other companies, is a narrative driver. The good news? Micron smacked market expectations in its most recent quarter Q3 FY26 : Revenue: $41.16 billion, up 74% compared to the sequentially preceding quarter, and up 346% compared to the year-ago period. Gross margin: 84.6%, up from 74.4% in the sequentially preceding quarter, and up from 37.7% in the year-ago period. Net income: $28.2 billion, up 105% compared to the sequentially preceding quarter, and up nearly 1,400% compared to the year-ago period. Even more, Micron anticipates revenue of $49.0 billion to $ 51.0 billion in its current quarter, with slightly higher gross margins ~86% . Shares of Micron shot higher following its report, rising around 10% this morning. A booming company in the historically volatile memory sector? What could possibly go wrong? Micron has a plan. Instead of merely riding demand booms and enduring demand busts https://www.federalreserve.gov/pubs/feds/2006/200644/200644pap.pdf that the memory market has long suffered, the company announced it has signed “16 strategic customer agreements, or SCAs, which we expect will fundamentally transform its business model.” Tell me more emphasis added Typically, these agreements have a five-year term, from calendar 2026 through the end of calendar 2030. … These SCAs are structured as take-or-pay agreements, with binding commitments to purchase specific volumes over this multi-year term. The largest agreements generally have a ceiling price for existing products at the current CQ2 calendar Q2 market price, and a floor price through the term of the agreement. … Fourteen of the 16SCAs that we have signed have a cumulative revenue at minimum price per our contracts of approximately $100 billionover the remaining agreement term.They also strengthen our long-term financial performance, margins and free cash flow expectations, with higher visibility and improved stability in our business performance. In simple terms, there is so much demand for the memory and storage products Micron sells that it has sufficient market power to force major customers to agree to long-term purchase commitments on favorable terms. For customers, the deals ensure supply; for Micron, they secure demand and margin , allowing the company to plan around stable sales over a multi-year period. Returning to why Micron’s earnings matter, investors were concerned about the durability of the current compute buildout and what a slowdown would mean for the software demand driving the data center rush. Micron crushed trailing expectations, guided above analyst estimates, and locked in major customers for years. Hardly the signs of a slowing market. - The above is why Micron was, and is still, a trillion-dollar company. Proof that the picks/shovels argument holds up no matter what particular boom we’re examining. With that out of the way, let’s chat inflation, AI gross margins, and good news from Crunchbase. To work — Alex 📈 Trending Up Getting left with the bag https://www.theguardian.com/us-news/2026/jun/25/jd-vance-iran-peace-deal … the best government money can buy https://www.politico.com/news/2026/06/24/white-house-helped-mark-zuckerberg-and-the-google-ceo-dodge-a-senate-grilling-00974497 … ballot box poison https://www.nytimes.com/2026/06/24/us/j-stuart-adams-utah-senate-data-center.html … startups smashing atoms https://x.com/isaiah p taylor/status/2069977748198785197 … supersonic drone jets https://x.com/twistartups/status/2069926141586968758 … AI regulation https://x.com/BenBrodyDC/status/2069871963930013881 ? … Inflation: Here in the ‘States, the personal consumption expenditures PCE rose 4.1% in May https://www.bea.gov/news/2026/personal-income-and-outlays-may-2026 , while core PCE less food and energy prices rose 3.4%. The figures largely matched expectations. We care about inflation because overall economic health matters, but from a tech-heavy perspective, rising prices matter for a second reason: tech stocks trade inversely to interest rates, and raising rates is a traditional tool to cool inflation. Expectations among investors are for rate increases at upcoming Fed moots https://www.cmegroup.com/markets/interest-rates/cme-fedwatch-tool.html , with no rate cuts from the current target rate until June 2027. Further boosting the chances of rate cuts are generally strong new data https://www.bea.gov/news/2026/gdp-third-estimate-industries-corporate-profits-state-gdp-and-state-personal-income-1st on first-quarter GDP growth https://www.bea.gov/news/2026/gdp-third-estimate-industries-corporate-profits-state-gdp-and-state-personal-income-1st and better-than-expected jobless claims https://www.reuters.com/business/us-weekly-jobless-claims-drop-more-than-expected-2026-06-25/ . - Ironically, tech is helping drive inflation, with the Journal reporting https://www.wsj.com/economy/the-data-center-boom-is-sparking-a-third-wave-of-inflation-926adc6e that a poll of economists found that “81% of respondents said the AI build-out will add to inflation over the next year.” We can see evidence of that inflation in rising prices at consumer giant Apple https://appleinsider.com/articles/26/06/25/apple-confirms-price-nikes-across-macs-ipads-and-more , which is grappling with rising memory prices see Micron above . AI gross margins? OpenAI’s improving gross margins matter https://www.cautiousoptimism.news/spacex-is-worth-more-than-amazon/ , and its upcoming chip project https://www.cautiousoptimism.news/why-openais-new-chip-will-help-it-go-faster/ could provide the AI lab and hyperscaler even better economics in the coming quarters. But what about Anthropic? Is it also seeing gross margin improvements? Yes, per Funda.ai https://funda.ai/ , a public equities analyst shop, which wrote the following https://fundaai.substack.com/p/deepllm-26h1-update-part-2-frontier : On customers, Anthropic’s stickiness shows in three data points: NDR 500% the same cohort’s next-year spend is 5×+ the prior year … Financially, Anthropic has turned cash-flow positive, with an FCF margin of ~15–20% and a gross margin of ~60–70%. One source does not a slam-dunk case make. But at the same time, did we think that Anthropic was losing money on Opus 4.8 inference? No. Thus, massive demand for Anthropic’s most capable models some data here https://openrouter.ai/anthropic should have a salubrious effect on its gross and net margins. We’ll need other sources on the FCF positivity point, but if OpenAI is seeing its margins scale from the 20s in 2024 to the 40s in 2025, why not the 60s in 2026? Crunchbase: Your humble servant owns shares in Crunchbase, having worked there for a few years shoutout to the Crunchbase News team . So it was to my personal benefit when, after expanding its predictive capabilities https://about.crunchbase.com/blog/market-insights — I interviewed Crunchbase CEO Jager McConnell about the initial rollout https://about.crunchbase.com/press/press-releases/crunchbase-declares-historical-data-dead-relaunches-as-an-ai-powered-predictions-engine-with-up-to-95-accuracy of data-based prognostications here https://www.youtube.com/watch?v=Ck9ddV o1KE — Crunchbase had this https://about.crunchbase.com/press/press-releases/crunchbase-introduces-market-insights-as-enterprise-demand-for-private-market-predictive-intelligence-drives-acv-growth to say earlier in June: The updated predictive capabilities launch comes on the heels of significant commercial momentum for Predictions and Insights in Crunchbase. In Q1 2026, Crunchbase saw average contract value ACV grow 900% year-over-year, driven by surging enterprise demand across financial services, investment firms and software/AI companies for trusted private market data to fuel AI transformation. The company’s two largest deals in its history closed in the last two quarters. … Looking ahead, Crunchbase’s 2026 roadmap includes model context protocol MCP connectors, a model-estimated current valuation signal, and more robust data coverage including revenue and headcount data, as well as expanded coverage of startups globally. Would you look at that I wondered if Crunchbase’s data and ability to chew it into something palatable would be a boon in the AI era. Seems that the answer is a somewhat delayed yes. Viva 📉 https://finance.yahoo.com/news/servicenow-pledges-1-5bn-investment-110000403.html Trending Down 📉 https://finance.yahoo.com/news/servicenow-pledges-1-5bn-investment-110000403.html The value of cryptocurrencies https://coinmarketcap.com/ … the German navy https://www.cnbc.com/2026/06/25/rheinmetall-defense-stocks-germany-f126.html ? … Microsoft’s quantum chops https://news.ycombinator.com/item?id=48661535 ? … being a weirdo https://www.wired.com/story/the-trump-white-house-is-over-anthropics-dario-amodei/ … Alibaba stock https://www.bloomberg.com/news/articles/2026-06-25/alibaba-drops-after-anthropic-accuses-firm-of-accessing-ai-model … paying humans https://www.ft.com/content/39251a31-4a9d-4870-b86c-dc6353d67fdd … thinking up good startup names https://www.wsj.com/tech/ai/anthropic-veterans-startup-seeks-to-help-scientists-develop-their-own-ai-09e2f3e5?st=CWXZSf&reflink=desktopwebshare permalink … Doing your own work: One way Chinese AI models quickly follow new, benchmark-setting AI models from the United States is by distilling prompting and learning from Anthropic and OpenAI models and applying that learning to the country’s own models. Anthropic made waves in February by directly accusing DeepSeek, Moonshot, and MiniMax of “using fraudulent accounts and proxy services to access Claude at scale while evading detection” to ‘extract capability’ from its work. We’re here again This time, Anthropic is pointing at Alibaba, telling Congress Bloomberg broke the story https://www.bloomberg.com/news/articles/2026-06-24/anthropic-accuses-alibaba-of-illicitly-accessing-its-ai-models , other publications confirmed it https://www.cnbc.com/2026/06/24/anthropic-alibaba-distillation-campaign.html that the Chinese tech giant executed “the largest known distillation attack on the company to date.” - It’s great that Chinese AI labs are working hard to provide low-cost intelligence to the world; IP matters, too. And cheating is for losers. Building for humans: Closing out the day, software design. Many tech leaders expect existing software products to ‘collapse’ into something far different from what they are today. In brief, the argument is that any software that is inherently CRUD-based Create, Read, Update, and Delete will see its human-facing UI dissolve into an interactive experience chatbot-mediated? or an agentic experience when the software talks to software rather than humans . So, we have our eyes out for startups building software purely for agents, as they may provide a useful signal about where the larger software market is heading. To wit: Seltz https://seltz.ai/ built its own “news index that’s faster and more accurate than any search API we tested, and that hands agents full, fresh documents instead of snippets.” To do so, the company had to build its own search engine https://x.com/antonio mallia/status/2069784143878132089 , a task it considers worth the work as “ traditional search was built for a human typing a few words and clicking a link but agents don’t search that way, and you can’t serve them by wrapping an engine that does.” Expect to see more of this.