Morning Edition
What gets rewritten without being struck down.
Three things happened in the last forty-eight hours that look unrelated and are not. The Court rewrote Section 2 of the Voting Rights Act without overturning it. Investors looked at the same number on four hyperscaler balance sheets and rewarded one of them and punished another, which is to say the AI capex thesis got rewritten in the time between earnings prints. And Vladimir Putin offered a May 9 ceasefire that is shaped exactly like the parade he wants to protect, which rewrites what a ceasefire is supposed to do.
What the three share is a method. Don’t strike a thing down; redefine it. The statute survives. The capex schedule survives. The word “ceasefire” survives. What changes is what each one means in practice, and the change is the story.
The paper is built around that thread. Front Page, AI, and World each take one of those rewrites at full size. The Locals offers a quieter version of the same: a Belmont assistant city manager moves to a neighboring city’s top job and inherits a different problem set without anyone announcing a doctrine. Food is the relief — a Hong Kong cafe being built in pop-up form by a Rich Table veteran who left a year ago and is letting the menu reveal itself in chef collaborations. The Edge is a physics paper showing that an antimatter atom can act like a wave, which is also a quiet rewrite of what we know about how matter behaves at the smallest scale where, at last, nobody is performing.
— The Editor
Section Two, gutted in name and not in body.
The Supreme Court did not overturn the Voting Rights Act yesterday. It rewrote what Section 2 requires plaintiffs to prove. The 2026 maps mostly hold. The 2028 maps do not.
The headline number from Louisiana v. Callais is 6-3, with Justice Alito writing for the majority and the three liberal justices dissenting. The headline ruling is that Louisiana’s two-majority-Black-district map, drawn after the Court itself blessed it in 2023, is now unconstitutional, and the original single-Black-district map is reinstated. The headline practical fact is that nothing happens to the 2026 House cycle, where most state filing deadlines have already passed.
The actual ruling is underneath. Alito’s majority did not strike Section 2. The opinion frames itself as “properly interpreting” a statute that has stood since 1965. What the new interpretation requires is that any race-conscious remedy in redistricting must rest on a “strong inference of intentional discrimination” — a bar that the old Section 2 jurisprudence specifically did not require, because the old jurisprudence treated discriminatory effect as sufficient evidence on its own. That is the rewrite. The statute persists; the four-decade understanding of how to use it does not.
What this means for 2028 is concrete. The ACLU estimates that roughly a quarter of the Congressional Black Caucus and roughly a tenth of the Hispanic Caucus could lose their seats once states in the South and Midwest begin redrawing maps that the old Section 2 jurisprudence would have killed. Texas’s “Big Beautiful Map” survived a parallel SCOTUS challenge the same morning, which tells you the new posture is not theoretical.
Read the dissent before you read the commentary. Justice Kagan’s argument — that the majority is using the language of statutory interpretation to do the work of constitutional repeal — is the argument the dissents in Shelby County made about Section 5 in 2013, and the next decade proved them right. The discipline of the moment is to take the rewrite seriously rather than wait for the formal overturn that may never come. The Court has shown, twice in twelve years, that it does not need one.
In Brief
Brent crosses $126.
Brent crude touched $126.41 overnight, a four-year high, after Trump told energy executives the Iran port blockade could run “for months” until Tehran agrees to a nuclear deal. CENTCOM says forty-one Iranian tankers holding sixty-nine million barrels are stranded. CNN
Q1 GDP advance prints at eight-thirty.
The Commerce Department’s first read on Q1 lands this morning. Atlanta Fed’s GDPNow tracks 1.2 percent; private-sector consensus runs 2.5 to 3.0. The wide spread sets up either a relief rally or a stagflation scare given $126 oil. Atlanta Fed
Apple after the close.
Apple reports Q2 2026 tonight against consensus of $109.7 billion in revenue and $1.94 EPS. The China tariff overhang and services growth are the live questions; the company is the Mag 7 outlier on AI capex because its inference is on-device. Apple Insider
ECB expected to hold.
Lagarde’s press conference at 8:45 ET; the deposit facility is expected to stay at 2.0 percent. The Iran-war “layer cake of shocks” — fuel costs, supply-chain pressure, currency stress — is the dominant variable, with markets now pricing three hikes in 2026 starting as early as June. CNBC
Selic surprise cut to 14.50.
Brazil’s central bank surprised consensus with a twenty-five-basis-point cut, the first cut by any major emerging-market central bank since the Iran war began. Mexico’s Q1 GDP today is expected at -0.5 percent quarter-over-quarter, the country’s first negative print since the pandemic. Rio Times
US Typhon batteries 100 miles south of Taiwan.
The US deployed Typhon mid-range missile systems to a Luzon Strait island during the Balikatan 2026 exercise; PLA Southern Theater Command sortied warships and aircraft in response. Taiwan-Philippines pork trade resumes May 14 after Taipei’s ASF-free recertification. USNI
FISA 702 reauthorized for three years.
The House passed the warrantless-query version of Section 702 reauthorization 235-191 Tuesday night, with Speaker Johnson attaching a CBDC ban to win conservative holdouts. Senate Majority Leader Thune has already killed the CBDC piece. The 702 extension itself moves cleanly. NPR
$100 million for a Mining Guard.
The Tshisekedi government announced a state-backed $100 million paramilitary force to protect cobalt and copper sites against artisanal incursion and smuggling, with US and UAE financing. Quietly the most significant EV-supply-chain story of the week. ACLED
TPLF reinstates the Tigray assembly.
The TPLF central committee voted to restore the Tigray Regional Assembly that had been suspended under the 2022 Pretoria peace agreement. Addis Ababa has not yet responded. The 2020-2022 war between the federal government and Tigrayan forces killed an estimated 600,000 people. ACLED
Khartoum bans imports to defend the pound.
Sudan’s transitional authority imposed a sweeping ban on industrial, consumer, and food imports after the pound fell roughly ten percent against the dollar in two weeks. Twenty civilians died Tuesday when a passenger boat overturned on the Nile in River Nile state. ACLED
The Locals
San Carlos hires its next city manager from one block north.
Kathy Kleinbaum, who was Belmont’s assistant city manager during the years that produced its current Housing Element and the Community Benefit framework, was named San Carlos’ next city manager by a council vote on Monday evening. The hire reads, as Peninsula hires often do, like succession planning across cities rather than a national search. Kleinbaum spent over twenty-five years in Bay Area public-sector roles, including stints in Belmont, San Mateo, and Los Altos, with a Stanford degree in Earth Systems and a Berkeley master’s in city and regional planning. Mayor Pranita Venkatesh called her “an outstanding choice” in a PublicCEO writeup that is otherwise the kind of bland personnel announcement these notices usually are.
The interesting thing is what she walks into. San Carlos is in the middle of a debate over a citywide RV-parking ban, a quiet Peninsula version of the kind of unhoused-policy fight that has consumed Berkeley and Oakland for a decade. Between January 2025 and April 2026, the city issued three hundred and forty-eight RV-related parking citations; only forty-four percent of those were paid. The current proposal would prohibit RV parking on any city street, leaving the question of where displaced residents go to be inherited by the next administrator. The Daily Journal covered the council debate.
Kleinbaum’s Belmont record is a specific kind. She helped the city land a Housing Element that has not yet been blown up by the state, ran a Business License Tax restructure that survived its critics, and built a Community Benefit framework that explicitly tied development approvals to local-services trade-offs. None of those are dramatic moves. They are the kind of careful institutional-capacity work that pays off five years later, when other cities are trying to figure out what their own framework should say. San Carlos is now the city that gets to find out what the second act looks like.
Food
Good Morning 96, in slow motion.
Gizela Ho left the Rich Table kitchen in November after roughly a decade running a line that earned the restaurant most of what restaurants in this city earn. What she announced this week, in a Standard interview with Sara Deseran, is what she was doing during the gap. She is opening a Hong Kong-style cafe called Good Morning 96 in 2027, named for the ubiquitous “good morning” face towels in cha chaan teng kitchens and for her grandfather’s 1996 Hong Kong restaurant.
The structure of the announcement is the interesting part. There is no brick-and-mortar yet. There is a calendar. May 3 brings a stall at the Bakers and Makers Fair at Nisei, with Hong Kong-Laotian sausage buns, mochi madeleines, a Hong Kong milk-tea crepe cake, lapcheong scones, and a black-sesame cookie. May 18 brings a collab dinner at Nisei and Angler for AAPI Heritage Month. June moves to Ox & Tiger. July to Prubechu. September to Peony Seafood. The cafe gets to exist for a year as a series of guest spots inside other people’s restaurants before it has to become its own room.
What the cha chaan teng tradition is, for anyone who has not eaten in one, is the specific Hong Kong synthesis where British colonial morning food (eggs, toast, milk tea) collides with southern Chinese street food (noodles, congee, char siu) in the postwar 1950s. The result is a cuisine that nobody planned. Pineapple buns with butter slabs. Macaroni soup with ham. French toast filled with peanut butter. Ho’s calendar, with its lapcheong scones and milk-tea crepe cakes, is operating in the same room — the diaspora cafe trying to serve not Hong Kong but the memory of Hong Kong, which is a different and harder dish.
FISA 702, on the move.
The thing the Speaker did Tuesday night is the kind of move that only registers as significant if you have been watching the seven-year fight over Section 702 of the Foreign Intelligence Surveillance Act. The House passed a three-year extension of 702, on a 235-191 vote, that does not include a warrant requirement for U.S.-person queries — the specific reform that civil-liberties groups have been pushing since the Carter Page case in 2017 and that conservative House members have, until recently, made into a bipartisan demand.
The reason the warrant requirement isn’t there is that Mike Johnson found a way to get the Freedom Caucus to vote for the warrantless version. He attached a separate provision to the bill: a ban on a U.S. central bank digital currency. That provision will not survive the Senate, where Majority Leader Thune has already told reporters the CBDC piece is dead on arrival. The 702 reauthorization itself, however, will pass. The conservative holdouts got a rhetorical win on something they care about. The Intelligence Community got the surveillance authority it actually wanted. The civil-liberties reformers, who needed the conservative coalition to hold, got nothing.
The mechanism here is worth slowing down on. Section 702 lets the NSA collect, without a warrant, the communications of foreign targets located outside the United States, even when those targets communicate with Americans. The “U.S.-person query” question is what happens after the collection: can FBI analysts run searches against the Americans’ end of those communications without a warrant? The current and now-extended answer is yes, with internal compliance procedures that the FBI’s own Inspector General has repeatedly found to be inadequate.
What just got reauthorized, in other words, is the part of the surveillance state that the post-Snowden reform consensus had specifically tried to bound. It got reauthorized for three years, which puts the next reform window in 2029, which is after the next Congress. Speaker Johnson did not announce this as a doctrinal shift. The doctrinal shift is the rewrite. The statute lives. What it now lets the government do is the question the next Inspector General report will have to answer.
The World
A ceasefire that is shaped like a parade.
The ninety-minute call between Vladimir Putin and Donald Trump on Tuesday produced one specific proposal, which Trump then repeated to reporters at the White House: a Russian-Ukrainian truce timed to May 9, Russia’s Victory Day. Putin’s framing, in a United24 readout, was that the truce would honor the eightieth anniversary of victory in the Great Patriotic War. Trump’s framing, in a Military Times piece filed Wednesday, was that he had “suggested a little bit of a ceasefire and I think he might do that.”
The catch, which Volodymyr Zelensky named within hours, is the date. May 9 is the day Russia holds its largest military parade of the year on Red Square, and it is the day Ukrainian drone strikes against parade-related infrastructure have, for two years, made the Kremlin most visibly uncomfortable. A unilateral pause that begins on May 8 and ends on May 10 is not a ceasefire in any operational sense. It is a one-day standdown of one specific Ukrainian capability, against which Russia would receive no reciprocal constraint, because no Russian capability is currently aimed at Red Square.
Read against what is actually happening in the field, the request is sharper. DeepState’s open-source mapping shows Russian forces advanced in or near ten Ukrainian settlements between April 21 and 28. A Russian mass drone strike on Odesa overnight Wednesday into Thursday injured at least eighteen civilians. The pause, in other words, is not being requested from a position of stalemate. It is being requested from a position of partial advance, by a leader whose major news of the week before had been the formal end of the Iran-mediated track of negotiations Trump had abandoned.
The thing Trump told the Kremlin, by accepting the framing publicly without naming the parade-protection function, is that the United States is willing to redefine “ceasefire” to mean “asymmetric pause.” The thing Zelensky said, on national television, is that he will not. The interesting question is not whether the May 9 truce happens. It almost certainly does not, in the form Putin wants. The interesting question is what the next U.S. proposal looks like, after the word has been redefined.
AI inside the catheter.
Abbott’s Ultreon 3.0 ships with the AI doing the read.
Tuesday’s Abbott press release announced FDA 510(k) clearance and CE Mark for Ultreon 3.0, the next generation of the company’s intracoronary OCT imaging platform. OCT — optical coherence tomography — is not a new modality. It has been the high-resolution imaging standard for percutaneous coronary intervention for years. The new claim is what the AI layer does inside the catheter lab.
The pitch is concrete. The pullback time is one second, down from the previous version. The AI does plaque-type assessment in real time during the procedure, and recommends stent landing zone and size before the operator commits. The system supports low-contrast and zero-contrast workflows — a meaningful operational claim, since roughly twenty-five percent of coronary-artery disease patients have impaired kidney function for whom contrast load is the binding limit on what you can safely do during a single intervention. Cardiovascular Business walks through the clinical trial register.
The deployments are arriving inline, not as autonomous reads.
The lesson for radiology operations isn’t that intracoronary OCT got faster. It’s the shape of where AI clearance is actually landing in 2026. Not autonomous chest reads. Not concurrent triage on standalone modalities. Inline workflow assistance, embedded in the procedure, where the value capture is procedure-time compression and contrast reduction rather than head-count reduction.
Avatar Medical’s 510(k) the same day — instant 3D processing pitched at IR, neuro, ENT, ortho, and oncology — fits the same pattern. Both are operating-room or procedure-room clearances, not reading-room clearances. The number to watch isn’t FDA clearances per quarter; it’s the share of those clearances that target the procedural pipeline versus the diagnostic pipeline. The first chart eats labor; the second reshapes it. Right now, in 2026, the procedural chart is the one moving.
For Stanford-side operational analytics, the practical follow is the contrast-volume number. If you can run an Ultreon 3.0 case at zero contrast on a CKD patient who would otherwise have been deferred, the throughput math changes. That is a measurable operational delta, and it is what the press release is implicitly selling.
Same number, opposite reactions.
The four hyperscalers will spend $725 billion in 2026, and the market just split on whether it’s worth it.
Wednesday night’s earnings prints were not the AI capex story. They were the moment the AI capex story stopped being unanimous. Alphabet raised its 2026 capex guide to $180-190 billion, with 2027 “significantly” higher, and the stock jumped seven percent after-hours. Meta raised its 2026 range to $125-145 billion, and the stock fell six. Microsoft told investors that capex would land near $190 billion for the year, including $25 billion of straight component-price inflation, and the stock went flat. Combined with Amazon’s $200 billion projection, the four hyperscalers are now tracking $725 billion in 2026 capex, roughly seventy-seven percent above 2025’s $410 billion.
The investor split is the story. Alphabet’s Q1 showed Cloud revenue up 63 percent year-over-year to roughly twenty billion, and management named enterprise AI as the cloud’s primary growth driver for the first time in any earnings call. Markets read that as the revenue line they could attach the capex to. Meta’s Q1 showed thirty-three percent revenue growth — the company’s fastest since 2021 — but no equivalent enterprise story; ads remain the engine, and the AI capex was framed as table stakes for staying competitive in recommendation systems and Ray-Ban Display production. Same number. Different attachment. Different reaction.
What this rewrites is the rising-tide thesis that has run from late 2024 through Q1 2026. For four quarters, every hyperscaler that raised capex got rewarded. The market was buying the spending as evidence of confidence. As of Wednesday night, the market is buying the spending as evidence only when it can see what the spending is going to produce.
The Column
The myths kept the names but lost the arrangement.
Sometime in the last six years, the project of retelling Greek myths from the women’s side stopped being a few breakout novels and became a publishing season. April 2026 is the loudest week of that season, and that is the column.
The breakouts you remember are Madeline Miller’s Circe in 2018 and Pat Barker’s The Silence of the Girls the same year. What followed was a slow build that didn’t quite cohere because it was happening in two adjacent rooms — literary fiction (Natalie Haynes, Jennifer Saint, Claire Heywood) and what publishers had started calling “romantasy” (Lore Olympus, the long shadow of Sarah J. Maas). Those two rooms are now visibly the same room, and the wall came down in a quiet way. April 2026 is when the floor plan reads as one space.
The evidence is the release calendar. Between April 22 and April 29, the U.S. shelves landed: a sapphic Medea-and-Atalanta-on-the-Argo reimagining; Sydney J. Shields’ An Arcane Study of Stars, dark academia with secret societies and after-dark deal-making — which is, transparently, the Faustus-and-Persephone material reorganized; and Suzanne Palmer’s Ode to the Half-Broken, which is not a Greek-myth retelling but is animated by the same instinct (a robot waking after collapse, named Be, in the abandoned Botanical Gardens, with a cyborg-dog companion). Andrew Liptak’s roundup for the month names this as one of the strongest months of 2026 for the sci-fi/fantasy field, and what’s running through it is consistent: women on the inside of stories that, in the originals, treated them as off-camera consequences.
The reason the column is interesting now and not three years ago is that the moves have stopped being defensive. The early Miller and Barker work was operating against a default: here is what the Iliad never told you. The 2026 work is operating after that default has been replaced. An Arcane Study of Stars doesn’t argue that women have been written out of the secret-society story. It assumes they’re in it and writes the next chapter. The Medea-Atalanta book doesn’t have to explain why two women on the Argo deserve a novel. It just goes.
That is what a publishing wave actually looks like when it has matured. The breakthrough novels did the work of changing the audience’s expectations. The post-breakthrough novels get to operate inside those expectations and try things. The genre-shaped questions about whether this is “really” myth retelling or “really” romantasy or “really” literary fiction become uninteresting at exactly the moment that the shelves stop sorting themselves into those rooms. The myths kept the names. They lost the arrangement. What’s left is the texture, which is the part the original storytellers were never going to give us anyway.
If you read one this season, the recommendation from this desk is the Medea-Atalanta. The Argo voyage in the original Apollonius is, formally, a series of cameos, and the modern retellings have the room to slow down inside any of them. The seam where two women’s interiorities meet on a ship full of mythic men is a structural problem that produces good prose as a side effect.
An antimatter atom acted like a wave.
A team in Tokyo got an “atom” of antimatter — a positron paired with an electron — to diffract through a grating, the way light or any quantum particle is supposed to. Nobody had managed it with antimatter before.
Positronium is what happens when a positron (the electron’s antimatter twin) and an electron orbit each other instead of annihilating immediately. They have about 142 nanoseconds together before the annihilation, which sounds like nothing and is actually plenty for a physics experiment if you have the right grating and the right detector. Tokyo University of Science, with collaborators, sent a beam of positronium through a transmission grating and measured the diffraction pattern on the other side. The pattern looked exactly the way quantum mechanics predicted matter waves should look. ScienceDaily’s writeup covers the Nature Communications paper.
The reason this matters past “neat trick” is that nobody actually knows whether antimatter falls. Specifically: when a positron sits in Earth’s gravity, does it accelerate downward at 9.8 meters per second squared, the way an ordinary electron does, or does it accelerate upward, the way some long-shot models suggest it might? CERN’s ALPHA-g experiment last year said the answer was “down, like normal matter,” with caveats. The positronium-diffraction technique gives experimentalists a different lever — a wave-interference handle on antimatter motion that should let you measure the gravitational acceleration with much higher precision than the bulk-trajectory experiments at CERN can.
What Tokyo demonstrated this week is the lever, not the answer. The point is that the lever exists. The next round of experiments will tell you whether antimatter’s gravitational behavior matches the rest of the universe or whether — somewhere far below the noise floor of every experiment that has ever been run — there is a difference. Either result rewrites a piece of the standard model. The technique is the news.
“Tomorrow is going to be wonderful because tonight I do not understand anything.”— Niels Bohr, recalled at Copenhagen
Morning Edition · For Paulo