Last month was the hottest March ever recorded for the contiguous United States — and not by a small margin. Federal data shows it exceeded previous records by the largest gap ever logged for any month, which is the kind of statistical anomaly that stops being reassuring to explain away. A forecast El Niño is expected to push temperatures higher still in the months ahead.

March being warm doesn’t make every summer extreme, and climate operates at scales that resist single-month narratives. But the margin matters. When records aren’t just broken but shattered, it suggests the baseline is shifting faster than the models anticipated, or that the models were right and we’re simply further along the curve than most public discourse has caught up to. Either interpretation leads to the same place.

A critical remote code execution vulnerability in Marimo — an open-source Python notebook tool used heavily in AI and data science workflows — was weaponized by attackers roughly nine hours after public disclosure. The flaw, tracked as CVE-2026-39987, allowed unauthenticated access. Within a working day of anyone knowing about it, someone was using it.

Nine hours is not an anomaly anymore; it’s a benchmark. The window between vulnerability disclosure and active exploitation has collapsed to the point where patch cycles that assume days or weeks of safety are simply wrong. For teams running notebook environments close to model infrastructure or internal data pipelines, this particular incident is worth treating as a drill. The CVE was patched; the lesson — that developer tooling is a high-value attack surface and exposure time is measured in hours — is ongoing.

Twenty-one hours of negotiations in Islamabad produced nothing. Vice President Vance, Steve Witkoff, and Jared Kushner sat across from senior Iranian officials — including the speaker of the Iranian Parliament — and walked away without an agreement. The sticking point, as Vance framed it, was Iran’s unwillingness to make a “fundamental commitment” not to develop a nuclear weapon or the means to quickly build one. That’s a reasonable bar to set. It’s also one Iran was always unlikely to clear under these conditions.

Calls are growing inside Congress for a formal investigation of Polymarket after another instance of anonymous traders making well-timed, well-sized bets on a major geopolitical event hours before it became public. The pattern is becoming hard to ignore: someone, somewhere, keeps knowing things before markets — or media — do.

Prediction markets have always carried this tension. Their theoretical value is as an information aggregation mechanism, surfacing the collective probability estimates of informed participants. The problem is that some participants are more informed than others, and not always because they’re better analysts. When bets arrive in size, in the right direction, in the hours before an event breaks, the question of whether markets are discovering truth or laundering inside information becomes unavoidable. Congress investigating Polymarket is probably not the most efficient path to an answer. But it may be the one that actually happens.

Microsoft, Google, Amazon and other major tech companies are locking in contracts with nuclear startups to secure reliable power for AI data centers. The rationale is straightforward: AI infrastructure runs continuously, consumes enormous amounts of electricity, and needs stable baseload power that solar and wind can’t reliably provide without storage that doesn’t yet exist at sufficient scale. Nuclear, which had been commercially stagnant for decades, is suddenly the destination of serious capital.

There’s a version of travel photography that involves light stands, permits, pre-dawn location scouting, and a pelican case full of glass. That’s not what this is. What this is — a woman moving along the marble terraces of the Bahá’í Gardens in Haifa, Sony mirrorless raised, backpack on, sunglasses pushed up into her hair because she forgot to take them off — is the other kind. The kind most people actually do. The kind that produces, against all odds, most of the photographs worth looking at.

The US Postal Service lost $9 billion in 2025 and has now warned it could exhaust its funds within twelve months. This is a structurally familiar story — mail volume declining, pension obligations compounding, the universal service mandate requiring delivery everywhere regardless of cost — but the twelve-month clock is new and specific.

USPS occupies a strange position in American infrastructure. It’s simultaneously indispensable to rural communities, pharmaceutical deliveries, ballot returns, and small-business shipping, and also chronically incapable of covering its own costs. Privatization discussions always die on the political vine because the math of profitable postal service requires abandoning the places that need it most. Subsidization discussions die because “bailing out the post office” is an easy campaign attack. The result is an institution that staggers forward indefinitely, absorbing losses, until a number like “twelve months” finally forces a decision that the political system has been deferring for a generation.

Andy Jassy disclosed in his annual shareholder letter that Amazon’s AI services at AWS are running at a $15 billion annual revenue rate. He also revealed that the company’s custom chip business — Graviton and Trainium — has crossed a $20 billion annualized run rate, roughly double what the company cited earlier this year.

Amazon stock jumped more than 5% on the news. The S&P closed at 6,824 and the Nasdaq at 22,822 on Thursday as the letter gave investors a concrete number to attach to years of capital expenditure narrative.

The Artemis II crew — Reid Wiseman, Victor Glover, Christina Koch, and Jeremy Hansen — are on their way home after becoming the first humans to fly around the Moon since Apollo 17 in 1972.

They’re the first humans to see parts of the lunar far side with the naked eye. That’s worth stopping to appreciate.

The mission was a crewed flyby — no landing — but the point was to prove the Orion spacecraft and Space Launch System could carry humans to lunar distance and back. By all accounts, it did.

Atlassian announced it is laying off approximately 10% of its global workforce — around 1,600 people — to redirect resources toward AI development and enterprise sales. The company estimates restructuring costs up to $236 million. It simultaneously replaced its CTO with two new AI-focused executives in the role.

This is now a recurring pattern in enterprise software. A company that built its reputation on collaborative tools announces it is dismantling some of its human workforce to build AI replacements for that workforce’s output. The framing is always “investment in the future.” The experience for the people being let go is something else.