The S&P 500 fell about 2 percent this week, and almost none of it was about earnings. It was about the cost of building the artificial-intelligence economy, the power, the silicon, the financing, a question the market had ignored for a year and repriced in five sessions. Micron’s blow-out memory quarter said the demand is real even as its sector was sold. And on Friday, four drones at the Strait of Hormuz revived the energy tail the market had just spent a fortnight pricing out.
Five numbers that frame the week: shares fell on the cost of the AI build-out, oil sits at pre-war levels before a Friday drone strike, the long end eased, and the fear gauge woke up. (Levels are verified Friday 26 June closes.)
“A repricing is not a verdict. It is the market changing its mind about the price of a story it still believes.”Anthony Rosenthal, Weekly Market Pulse, June 2026
TL;DR
Markets in brief. The S&P 500 closed Friday 26 June at 7,354.02, down about 2 percent on the week and up 7.4 percent for the year, but holding well above its June lows even through the chip-stock rout, the tell of a market still climbing a wall of worry. The 10-year Treasury yield, which sets the cost of mortgages and business loans, eased to 4.40 percent as oil fell, while the rate-sensitive two-year sat near 4.09 percent, leaving the curve re-steepened. WTI crude (West Texas Intermediate, the main US oil benchmark) closed at 69.23, down nearly 10 percent on the week and at pre-war levels, before Friday’s Hormuz drone strike put a fresh bid under it. Gold eased to about 4,079 dollars, down 6 percent on the year; silver was hit hardest of all, down about 16 percent year-to-date. The VIX, Wall Street’s fear gauge, rose 15 percent on the week to 18.89, still below 20 but no longer calm.
The week’s hinge variables. Three conditions decide where this goes next. First, whether the VIX closes above 20 and holds three sessions, which would turn the AI-cost wobble into a volatility regime rather than a one-week flush. Second, whether Friday’s Hormuz drone strike escalates into a confirmed supply event that puts WTI back above 80 dollars and the 10-year back above 4.55 percent. Third, whether high-yield credit spreads stay below 350 basis points, which would keep the equity repricing out of the plumbing. Each is taken up in the Analytical Takeaway.
The crash gauge ticks up to 27.5 but stays in the benign band, and every point of the rise is the fear gauge waking up on the tech rout, not the credit market, which stayed calm.
Three ways the next ninety days could break, a healthy repricing of expensive equities is the base case, with a Hormuz-plus-capex unwind the tail.
| Scenario | Probability | Trigger | 2Y | 10Y | Equity impact |
|---|---|---|---|---|---|
| Healthy Repricing | 50% | Dip-buyers return, AI-capex names stabilise; VIX falls back below 17 and breadth holds 60 percent | ~4.1% | ~4.4% | Expensive multiples cool, then the wall of worry holds |
| Choppy AI-cost digestion | 35% | Rangebound, rotation under the surface; VIX 18-24, no credit stress, the curve stays steep | ~4.1% | ~4.4% | Choppy, narrow, no clean trend |
| Hormuz re-escalation + AI unwind | 15% | Friday’s drone strike escalates into a confirmed Strait disruption AND the fear gauge breaks above 28 | ~4.0% | >4.55% | Broad risk-off; energy tail and crowded tech unwind together |
Last week’s tell, scored first. A week ago the tell was simple: if the 2-year/10-year spread holds above zero into Friday, the rate market has confirmed it is pricing a slowdown, not a hike. It held, the curve stayed re-steepened all week, ending around plus 31 basis points even as the 10-year itself eased. The slowdown read is confirmed by the bond market. The tell fired correct.
The honest framing of this week is in the hero quote: the market changed its mind about the price of a story it still believes. Nothing in the AI-infrastructure thesis broke. Micron’s results said demand is, if anything, stronger than priced. What changed is that investors started doing the arithmetic on the cost side of the build-out and decided the multiple had run ahead of the bill.
That distinction matters for positioning. A story that is breaking, an earnings miss, demand evaporating, fraud, you sell. A story that is repricing, the market re-rating the cost of capital for an intact thesis, you treat more carefully: you let the froth come out and you watch which names hold their fundamental floor. The chip-stock rout this week was indiscriminate, it sold the companies whose demand Micron just confirmed alongside the ones running on narrative. That is what a repricing looks like, and it is usually a worse moment to sell than to buy with discipline. The cleaner competing read is worth stating plainly: this could be a positioning unwind in the most-crowded semis rather than a repricing of the thesis itself. The tell that separates them is whether the selling persists once the crowding clears, or broadens into names that were never crowded, the first is froth coming out, the second would be the thesis itself in question.
The cross-asset tape is internally consistent for the first time in a month. Equities down, oil down, the long end down, gold down, crypto down, that is not panic, which bids gold and the dollar. It is a de-risking: money coming off the most-crowded trades, mega-cap tech and the AI complex, without rushing into a haven. The re-steepened curve says the bond market agrees: this is a growth-and-cost story, not an inflation-and-Fed story. The one thing that could turn the de-risking into something with a haven bid is Hormuz, which is why the energy tail sits at the top of the watch list.
The expiry date on resilience. One seasoned allocator framed the US position this week with a precision worth borrowing: the resilience is real, and it has an expiry date. The historical record on oil shocks of this size is unusually clean. Going back to 1986, when a shock of this magnitude resolves within about three months, equities recover and carry on, positive returns in the large majority of cases. When it persists beyond six months, the record turns hard the other way: a median S&P 500 loss of around 15 percent over the following year, and not one positive instance in the data. The S&P has recovered every point it lost since the conflict flared this spring. That is the good news. The clock is the bad news: the six-month threshold the record flags arrives toward the early autumn, and the market is pricing the recovery, not the scenario in which the Strait does not reopen before it. (Data: Goldman Sachs Investment Strategy Group, June 2026; framework recreated by Weekly Market Pulse.)
The mirror in the AI accounts. A second thread deserves more scrutiny than it has had. Under current accounting rules, a company that owns shares in an AI start-up can record a profit each time that start-up raises money at a higher valuation: no shares sold, no cash changing hands, and the paper gain flows straight into reported earnings. It is entirely legal. It gets harder to read when the company booking the gain is also the one selling the start-up the computing power it needs to survive, the investor’s own spending helps create the demand that lifts the valuation it then marks its investment up against. Stack that across a sector where every funding round is public, and some portion of the AI returns being reported looks less like an independent market verdict and more like a mirror. This is not an allegation of wrongdoing; it is a measurement question, and an under-asked one, which is why it also tempers how much of the AI earnings miracle should be read as pure demand.
The “so what” for a real person. If you hold a broad equity fund, this week cost you about 2 percent and told you very little about the next year. The thing to watch is not the AI headlines, it is whether the cost repricing stays contained to equity multiples or starts showing up in credit spreads. It has not yet; high-yield is calm. If credit stays calm, this is a multiple wobble. If high-yield spreads start widening through 350 basis points (hundredths of a percentage point), the repricing has found its way into the plumbing, and that is a different conversation. And there is a constructive reading worth keeping in view: a market repricing the cost of an intact story, with the demand under it confirmed by Micron and US earnings resilience holding, is repricing price, not progress.
An amber tail, named in both directions: if Iran follows Friday’s drone strike with a confirmed closure event in the Strait, expect WTI back above 80 dollars within days and a 30-to-50 basis-point back-up in the 10-year regardless of the growth story, the energy-inflation trade the market just buried would reassert itself fast. And the standing corrective: the week’s clearest negative catalyst, the chip rout, did not produce a new reaction low in the S&P, which sits well above its June lows. The wall-of-worry regime is still live, so this week’s view is neutral-with-a-named-tail, not bearish.
Where I could be wrong (the standing self-check): the comfortable read this week is “repricing, not break.” It is wrong if the chip selloff is the first domino of a real capex retreat rather than a cost wobble, and the tell is a single hyperscaler trimming its data-centre budget. The growth-and-cost framing is wrong if the two-year climbs and the curve flattens back toward inflation-and-Fed. I score this view against both.
S&P down about 2 percent; the Nasdaq’s fifth straight losing session; Micron beat hard on AI memory; the OpenAI IPO reportedly delayed; oil down nearly 10 percent to 69.23 then a Friday Hormuz drone strike; 10-year 4.40 percent; VIX 18.89. The crash gauge ticked up to 27.5.
The selloff is a cost-side repricing of an intact thesis, not a demand break. The de-risking is broad and haven-less, so it is positioning, not fear. Credit staying calm makes this a wobble, not a crack. The GS ISG record warns the resilience is conditional on the Strait reopening before the autumn.
A confirmed Hormuz closure fires the energy tail. A high-yield spread move through 350bps means the repricing has entered credit. A second AI-capex name guiding capex down would turn a multiple story into an earnings story, at which point the read changes.
Who they are. NewLimit is the kind of company that only makes sense if you believe a specific, unfashionable thing: that ageing is not a one-way street but a software problem written in the wrong order, and that the order can be rewritten. It was founded in 2021 by Brian Armstrong, the same Brian Armstrong who built Coinbase, together with the investor Blake Byers and the scientist Jacob Kimmel, who runs it. For four years it did the least glamorous thing a biotech can do: it stayed quiet and generated data. This month it stopped being quiet. On 2 June it announced a 435-million-dollar funding round led by Founders Fund at a reported 3.1-billion-dollar valuation, a number that says serious money has decided the data is real.
The problem. Every cell in your body carries the same genetic code, but a liver cell and a skin cell read completely different parts of it. That reading instruction, which genes are switched on and which are off, is the epigenome, and as you age it degrades, like a library where the index slowly scrambles. NewLimit’s bet, building on the Nobel-adjacent work on cellular reprogramming, is that you can partially reset that index, make an old cell read its genes the way a young one does, without erasing the cell’s identity, so a liver cell stays a liver cell. Do that safely in a living human and you have not cured one disease; you have addressed the thing underneath many of them.
The moment of real uncertainty, and the stake. The uncertainty here is not commercial, it is existential to the thesis, and it arrives at exactly one moment: the first time the in-dish result has to survive contact with a living liver. NewLimit has shown age-reversal signatures in human cells in a dish. The whole thing rests on whether that replicates in a person, and on whether partial reprogramming can be dialled in precisely enough to rejuvenate without tipping a cell toward the uncontrolled growth that is the definition of cancer. That is the knife-edge: the same mechanism that makes a cell young is one dial-turn from the mechanism that makes a cell a tumour. Armstrong has put his Coinbase fortune and his reputation behind a problem where the failure mode is not “it does not work” but “it works and then it does something you cannot allow.” That is a different kind of courage from building an exchange.
What it teaches. The lesson is about how to think about a binary. NewLimit is not a spreadsheet exercise; it is an option on a single replication event. The discipline, the same one this publication applies to every call it makes, is to write down before the data the specific result that would make you wrong, and to refuse to move the goalposts afterwards. For NewLimit that line is explicit: the first-in-human liver data fails to replicate the in-dish age-reversal, or a safety signal of uncontrolled growth appears. If either fires, the story is over, full stop, no “but the science is still exciting.” We can’t invest, and we don’t have all the facts; we run it because it is one of the most genuinely interesting things happening at the frontier of medicine.
The stratification question. If NewLimit works, the first people to access cellular rejuvenation will be the people who can pay for a first-in-class therapy, which means a technology that addresses ageing itself could, at launch, widen the most fundamental inequality there is: how long you get to live. The governance problem is not hypothetical and not distant; it is the default outcome unless someone designs against it. A serious observer of this category has to hold two things at once: the genuine human upside of compressing late-life illness, and the real risk that longevity becomes the ultimate positional good. The WMP names it because a case study that does not is a press release.
The number that framed the week was not an earnings figure. It was the multiple. The S&P 500 gave back roughly 2 percent to close at 7,354, and the Nasdaq fell for a fifth straight session, but corporate results, where they landed, were good.
Micron, the hardest numbers of the week. Micron reported a blow-out fiscal third quarter on 24 June: revenue more than tripled year on year and gross margins cleared 81 percent, with its high-bandwidth-memory line, the stacked chips that sit next to the GPUs inside AI accelerators, already sold out under multi-year contracts. AI-memory demand and pricing power were confirmed in the most concrete figures the week produced. FedEx beat on 23 June, with adjusted earnings of 6.31 dollars a share and revenue up 13 percent to 25.0 billion dollars, a quiet but real signal that goods are still moving. Darden beat too, though Olive Garden’s same-store growth softened, a small consumer tell worth filing.
So why did the market fall? Because the thing that sold off was not earnings. It was the price of the build-out. A reported delay to the OpenAI public listing landed in the middle of a week when investors had already started questioning the bill for AI data centres, the hundreds of billions of dollars of capital expenditure behind them. Chip stocks led the decline because they are the purest expression of that bill. Apple and Microsoft announcing iPhone and Xbox price increases on 25 June added a faint second note, margins and the consumer in the same breath.
The honest summary. A week where the companies did well and the market did not, because the market spent five sessions re-rating how much the future costs rather than whether it arrives.
The plumbing is calm and the mood is not, this week’s risk was in equity multiples and in how the AI build-out is being financed, not in credit.
A composite score of 27.5 says the same thing in plain language: the crash gauge is still in its benign band, but it ticked up, and every point of the rise came from equity-market nerves, the VIX, not from the credit market, which stayed calm. In practice this was a week to let the most-crowded trades cool, not a week to reach for the exits. The single number that would change this read is a high-yield credit spread pushing through 350 basis points. It has not. Until it does, treat the move as a repricing of expensive equities and a fair question about how the AI build-out is financed, not a crack in the system.
The speed story this week is the speed of a repricing. It took the market roughly five trading sessions to move from “AI build-out is the only trade” to “wait, what does the build-out cost?”, a question that had been answerable in public filings for a year. The information was not new. The willingness to act on it was. That gap, between when a fact is knowable and when the market decides to price it, is where most of the edge in this job lives.
This takes about four minutes. Paste the following into Claude: “Here are the three biggest US AI data-centre capital-expenditure plans announced this year. For each, estimate the annual electricity demand in terawatt-hours and compare it to the total power consumption of a mid-sized country. Show your working and flag where you’re uncertain.”
What it teaches you: the point is not the exact number, it is to feel, in your own hands, how quickly you can turn a vague worry (“AI uses a lot of power”) into a comparison you can actually reason about. When Anthony ran a version of this, the striking part was not the size of the demand but how uncertain the public figures are, which is exactly why the market can ignore a cost for a year and then reprice it in a week. The reader who can build that comparison themselves is no longer at the mercy of the headline. That is the whole game: not faster answers, but the ability to interrogate the question yourself.
Two fronts moved this week, and they pull in opposite directions on the same asset: oil.
Hormuz, the ceasefire the market priced out came back. For two weeks the dominant trade was de-escalation: WTI fell to pre-war levels near 70 dollars as the market treated the Iran confrontation as effectively over. It is not. On Friday 26 June, Iran reportedly fired four one-way drones at the Strait of Hormuz, one striking a large cargo ship, a breach of the 60-day ceasefire extension. The analytical point is the one this publication has held for eighteen weeks: signature, not rhetoric. The Geneva technical talks collapsed on 18-19 June; the much-reported memorandum of understanding was never signed. So the market that priced oil down toward 70 dollars was pricing a peace that does not exist on paper. Friday’s drones are a reminder that the energy tail was dormant, not dead. The insurance signal, force-majeure clauses on Gulf shipping, remains the most reliable leading indicator; watch whether it reactivates.
Rare earths, the dispute widens. On 22 June China added MP Materials and USA Rare Earth, the two largest US federal rare-earth bets, to its export-control list, retaliation for the Pentagon’s roughly 80-name critical-minerals listing. The immediate price reaction was muted, and the reason is instructive: the curbs restrict what these firms can import from China, the processing inputs and equipment, not what they can sell. So the curbs harden the domestic-supply thesis rather than breaking it, making the case for a Western rare-earth supply chain more urgent, not less. The defence-drone-rare-earth complex is now formally inside the US-China dispute. That is a structural development, and it is why both names stay on the watch list even with no fresh price catalyst.
The “so what”. The same oil price sits between a structural ceiling, the post-Hormuz bypass infrastructure and demand destruction, and a live geopolitical floor, a Strait that can be threatened with four cheap drones on a Friday afternoon. That two-sided risk is exactly why the energy directional view stays amber, and why a confirmed closure event is the single most important thing to watch into next week.
The reflex this week was to ask what is wrong with the AI trade. The more useful question is where the disciplined money is going instead, and the answer is unglamorous on purpose.
A senior credit investor at one of the world’s largest alternative asset managers described their current flagship private-equity fund this week in terms that would have sounded quaint two years ago: it is buying cash-generative, operationally real businesses at six to eight times EBITDA (earnings before interest, tax, depreciation and amortisation, what a company earns from its core operations before debt and tax). For context, the 2021-22 peak saw funds paying fifteen to twenty-five times for growth stories. Most private-equity returns in that era came not from improving the business but from buying it at twelve times and selling it to someone willing to pay eighteen, “multiple expansion,” which only works while the next buyer stays more optimistic than you were. Six-to-eight-times entry needs none of that: the return comes from the business doing what it already does, with some debt paid down over time. In the vocabulary of this publication’s Repricing thesis, this is the other side of the trade, the moment disciplined capital that sat out the speculation finally finds prices it recognises.
The same migration shows up at the yield-seeking end, in a corner most Western allocators never see, and one open only to a specific group: Indians living abroad, through the foreign-currency deposit accounts reserved for them. Certain Indian banks are reportedly structuring a leveraged facility on top: the affiliated lender advances the loan in the same foreign currency against the depositor’s deposit, the whole amount goes into a higher-paying foreign-currency deposit at an Indian bank, and the depositor keeps the spread between that deposit rate and the lower cost of the borrowing. The headline figures deserve care, around 15 percent at up to nineteen times leverage is the aggressive ceiling, not the norm; the more typical structure is closer to nine times for roughly 14 percent, and the risk and reward only look sensible below about nine times, before the leverage swamps the cushion. Two things are worth saying plainly. The bank is the surest winner here: the loans are fully secured against the deposit and made with recourse, so it is the depositor, not the bank, who carries the leverage risk, which is exactly why the one lever that matters is negotiating hard on the spread the bank charges. And Western regulation forbids most of this structure because leverage amplifies losses as efficiently as gains; India’s framework treats these non-resident deposits differently, and the government wants the dollars coming in. The point is not the product, which is closed to almost everyone reading this. It is the distance capital is now travelling, and the leverage it is taking, to find a number it used to get at home.
A major alternatives manager is entering at 6 to 8 times EBITDA, against 15 to 25 times at the 2021-22 peak; leveraged foreign-currency deposit structures (non-resident Indians only; typically about 9 times for roughly 14 percent, with 19 times and 15 percent the aggressive ceiling) are being offered, with the bank fully secured and the spread the depositor’s key lever.
Disciplined capital is deploying because the Repricing has finally produced prices it recognises; yield-seeking capital is migrating far outside Western fixed income, and taking real leverage, to find its number.
The rupee leg is not the depositor’s risk to carry: an FCNR deposit is held in foreign currency, and the rupee hedge sits with the bank and the RBI’s swap facility, not the non-resident. Nor is a yield spike the immediate worry, the loan and the asset are matched in duration and held to maturity, so a price wobble does not force a margin call. The genuine exposure is convertibility: a sovereign move to restrict repatriation would leave the borrower still owing a hard-currency loan while the asset sits trapped inside India. Remote and largely theoretical, but it is the tail a geopolitical shock could open, and it is the borrower who wears it; separately, a credit-cycle turn would test whether the 6-to-8-times entry multiples were early rather than cheap.
In December 2025 the New England Journal of Medicine published something that got far less attention than it warranted: a prime-editing therapy, Prime Medicine’s PM359, that restored the missing immune function in patients with chronic granulomatous disease, a severe inherited immune disorder that until now required a bone-marrow transplant. The corrected cells engrafted promptly and the function held through follow-up, with no therapy-attributable adverse events, the first published clinical proof that prime editing works in people. Prime editing is the generation of genetic medicine after CRISPR: it rewrites single letters of DNA without the double-strand breaks that make CRISPR unpredictable in the clinic. Think of CRISPR as cutting a sentence out with scissors and hoping the cell stitches it back correctly, and prime editing as using the track-changes tool to correct one letter. A published, durable functional correction is the proof-of-concept the whole field has been waiting for.
The capital has noticed. When two pharmaceutical giants get into a bidding war for a precision-biology company, the message is unmistakable. Metsera was acquired in late 2025 in a transaction valued at up to 10 billion dollars, alongside Bristol Myers Squibb’s 1.5-billion-dollar purchase of Orbital Therapeutics. The industry has decided precision biology is the next competitive arena, and is willing to pay to secure the capability. Conversations this week with people building at the frontier of life-sciences venture make the same point from the other side: the exits are starting to validate the science faster than the funding cycle assumed. This is the kind of progress the week’s oil-and-AI anxiety obscures, which is precisely why it belongs here. The science cited is public and peer-reviewed; no private fund or position is named or implied.
Each week the WMP scores the central economic argument about artificial intelligence as a contest between two forces. The Displacer is the case that AI substitutes for human labour, concentrating the gains in capital and eventually destroying the spending power the economy runs on. The Augmenter is the case that AI raises human productivity, expands output, and spreads the gains broadly. Both sides agree AI is powerful. They disagree about whether it is a demand shock or a supply shock.
This week, four pillars, one point each:
Running total, the Augmenter 16, the Displacer 4. The Augmenter takes three pillars to the Displacer’s one for the fourth week running, but the hook is the shape, not the margin: the sell-off itself is the Augmenter’s argument. For four months the Displacer’s strongest card has been the relentless capex surge; this week the market looked at the bill for that surge and flinched, which is precisely the Augmenter’s claim that the compute-cost ceiling binds before mass displacement does. The Displacer keeps its grip on the economic-shock pillar a third straight week, but it lost the argument it usually wins. What would flip the score: a single white-collar displacement print naming AI as the structural cause, or a visible consumption-gap opening in the spending data, takes the week for the Displacer; the Augmenter holds while the labour market keeps growing and the compute-cost ceiling binds.
The ERDR framework tracks twelve income strategies that aim to earn an equity-like return for taking on debt-like risk. This is an even week, so there is no deep dive; the Standing Dashboard refreshing all twelve follows. The terms, briefly: a spread is the extra yield a strategy pays over a safe government bond, in basis points (hundredths of a percentage point); investment grade means the safest tier of corporate borrowers; and a floating-rate strategy is one whose income rises automatically when short-term interest rates rise.
| Strategy | Indicative Yield | Spread vs IG | WoW | Action |
|---|---|---|---|---|
| 1. Active income fund + Lombard | 6.9 to 7.7% | +300 to 380bps | Cost eased with the front end | Watch |
| 2. IG / split-rated CLO tranches | 6.3% | +220 to 260bps | Floating coupon drifts with rates | Hold |
| 3. Listed infrastructure debt/equity | 5.7% | +175bps | Firm on power demand + lower long end | Add |
| 4. Business development companies | 10.8% | +560bps | Floating yields ease; watch credit quality | Hold |
| 5. Agency mortgage REITs | 13.1% | +150bps (asset OAS) | Helped by the long-end rally | Watch |
| 6. Senior secured leveraged loans | 8.5% | +420bps | Floating-rate, resilient; yield drifts down | Hold |
| 7. Preferred shares / hybrids | 7.2% | +295bps | Firmer as the long end eased | Add |
| 8. Real asset royalties | 6.6% | +255bps | Soft as oil fell further | Hold |
| 9. EM hard-currency sovereign carry | 7.9% | +385bps | Steady; softer dollar a help | Hold |
| 10. High-yield municipal bonds | 6.1% (tax-free) | +275bps | Firmer with lower long rates | Add |
| 11. Private credit direct lending | 11.0% | +575bps | Spreads holding; liquidity-turn caution | Hold |
| 12. Trade & supply-chain finance | 8.9% | +450bps | Short-tenor, defensive | Add |
Each strategy is explained in full when it is the week’s Deep Dive.
With the 10-year easing to 4.40 percent and high-yield spreads calm around 278 basis points, the tilt this week is subtle. The floating-rate strategies, senior secured loans, BDCs, direct lending, see their headline yields drift down with the front end, while the fixed-spread, longer-duration strategies, listed infrastructure debt, preferred and hybrid capital, high-yield munis, get a modest tailwind from the lower long end, which is why they move from Hold toward Add. The standing caution remains private credit direct lending: a Howell liquidity cycle near its peak is exactly the environment in which direct-lending marks are slowest to reflect stress, so the calm is partly an artefact of infrequent marking. No thesis changed this week; spreads moved with rates, not with risk.
What I am watching and why, not a recommendation to buy or sell. A scoring week: no new name, but a four-week call comes due, and the running record is reported with the misses kept in.
The public scorecard. Four weeks ago this publication put Arista (NYSE: ANET) on the radar at 147 dollars with a simple thesis: the market was underpricing its position as the networking layer of the AI build-out. This Friday closed its four-week tactical window at 165.45, up 12.6 percent, a tactical win. That is the timing question answered; the bigger analytical question, is the thesis right, does not close until 28 August. Honesty in both directions: the running tactical scorecard now reads thirteen of nineteen correct, about sixty-eight percent, with six misses, and the misses get named with the winners. The two current laggards are Venture Global, down roughly 5 percent from entry with the LNG thesis tactically early, and MP Materials, down about 13 percent, caught in the rare-earth crossfire. No thesis-horizon verdict has yet been recorded; the first two close on 18 July, on Bloom Energy and Freeport. The track record is a number that can go down, and it is published every four weeks precisely so it can.
No new listed name this week, and that is the discipline working. USA Rare Earth’s Brazilian antitrust decision, the CADE ruling on the Serra Verde acquisition, which would have been a genuine company-specific catalyst, did not arrive; the investigation remains open with the deal still guided to close in the third quarter. Because there is no named, seven-day, company-specific event on any active name, nothing was promoted from Portfolio Watch into the main section. A favourable price move is not a catalyst. That is the rule, and this week it kept a name off the page.
Arista Networks (NYSE: ANET) · entry $147.00 (Wk 20) · four-week tactical CLOSED 26 Jun at $165.45, +12.6% · thesis horizon 28 Aug
SpaceX (SPCX) surfaced this week as a contrarian-frame candidate, a bear case on the orbital-data-centre pillar of its valuation, but it fails the media-silence screen as the largest IPO in history and is logged as a tracked external call, not promoted. The eight active calls, Bloom Energy, Talen Energy, Arista, USA Rare Earth, Freeport-McMoRan, MP Materials, Venture Global and YPF, are all tracked in Portfolio Watch below.
There is a particular comedy in a week where the companies reported good news and the market sold them anyway. Micron told the world its AI-memory business is on fire; its sector spent the week being doused. It is the financial equivalent of being congratulated and fined in the same letter.
The deeper joke is on all of us who spent a year being told the AI trade had no discipline, only to panic the moment the discipline showed up. The market wanted someone to ask what the build-out costs. Someone did. Everyone ran. Your columnist, an Arsenal supporter and therefore a lifelong student of paying a great deal now for a future that may or may not arrive, found the whole thing oddly familiar.
The market loved AI’s bright story,
Then totted the bill for the glory.
Good earnings still came,
But it sold them the same,
Repriced, not yet purgatory.
Three things will tell us whether this week was a wobble or a turn. First, the next hyperscaler capex signal: any hint of a data-centre budget being trimmed turns a cost story into an earnings story. Second, the Strait of Hormuz: whether Friday’s drones were a gesture or the start of something that puts oil back above 80 dollars. Third, high-yield credit spreads: whether the equity repricing stays out of the plumbing. If all three move the wrong way at once, capex cut, oil up, spreads wider, the de-risking becomes a de-rating and the benign band on the crash gauge will not hold. If they diverge, this was the market growing up. We will know more in seven days.
2026 Scoreboard
25 assets ranked by year-to-date return · Baselines locked 1 January 2026 · Week 24 · 26 June 2026 · Verified Friday 26 June closes; Nasdaq 100, LQD and MSCI EM carried with a flag where the data feed had a gap (see exception report); Baltic Dry from Trading Economics (26 Jun); MSCI EM derived from EEM × locked ratio; crypto on a 24/7 basis
The five Scoreboard asset classes are still pulling in different directions, the S&P and oil up on the year, bonds and gold down, credit calm, a directional, if moderate, reading of the Repricing thesis at the mid-point.
2026 YTD Performance, All 25 Assets, Week 24
| Rank | Asset | 1 Jan 2026 Baseline | Week 24 Close (26 Jun) | YTD % |
|---|---|---|---|---|
| 1 | Nikkei 225 | 51,830 | 72,366.34 | +39.62% |
| 2 | Baltic Dry Index | 1,882 | 2,524 | +34.11% |
| 3 | USD/TRY | 35.40 | 46.55 | +31.49% |
| 4 | Russell 2000 | 2,481.91 | 3,007.86 | +21.19% |
| 5 | MSCI EM | 1,595.20 | 1,907.68 | +19.59% |
| 6 | Nasdaq 100 | 25,200.50 | 29,440.32 | +16.82% |
| 7 | WTI Crude | $63.20 | $69.23 | +9.54% |
| 8 | Euro Stoxx 50 | 5,740.15 | 6,267.53 | +9.19% |
| 9 | Copper | $5.682 | $6.14 | +8.09% |
| 10 | S&P 500 | 6,845.50 | 7,354.02 | +7.43% |
| 11 | Swiss SMI | 13,248.10 | 14,231.96 | +7.43% |
| 12 | FTSE 100 | 9,948.30 | 10,529.90 | +5.85% |
| 13 | HYG | $78.15 | $79.88 | +2.21% |
| 14 | DAX | 24,540.20 | 24,994.83 | +1.85% |
| 15 | LQD | $109.02 | $109.50 | +0.44% |
| 16 | Nifty 50 | 24,420 | 24,056 | -1.49% |
| 17 | AGG | $102.15 | $99.36 | -2.73% |
| 18 | USD/ZAR | 17.55 | 16.50 | -5.97% |
| 19 | Gold | $4,341.10 | $4,078.70 | -6.04% |
| 20 | TLT | $94.27 | $87.29 | -7.40% |
| 21 | Natural Gas | 3.514 | 3.231 | -8.05% |
| 22 | Hang Seng | 26,340 | 23,076.91 | -12.39% |
| 23 | Silver | $70.61 | $59.22 | -16.13% |
| 24 | Bitcoin | $87,850 | $60,016 | -31.68% |
| 25 | Ethereum | $2,967 | $1,577 | -46.86% |
Verified Friday 26 June closes, with the Scoreboard rendered live from the Supabase record for edition 24. Nasdaq 100, LQD and MSCI EM are carried with a flag where the price feed had a same-day gap (re-verified on the Monday scoring run); MSCI EM is derived from the EEM ETF close multiplied by the locked index ratio (28.367); Baltic Dry is the named Trading Economics reading (2,524). All baselines locked 1 January 2026.
Every company that has appeared in On the Radar is tracked here until its formal score date. A company moves from On the Radar to this appendix when there is no fresh catalyst that week, the analytical call is intact, but there is nothing new to add.
| Company | Entry | Week | Current Close | Return | Original Thesis | Score Date |
|---|---|---|---|---|---|---|
| Bloom Energy (NYSE: BE) | $207.10 | Wk 14 | ~$271.09 | +30.9% | On-site fuel-cell power for AI data centres, deployable in months versus multi-year grid queues | Thesis score 18 Jul. |
| Six weeks in and still up a third, but down about 12 percent on the day on profit-taking from its $309 high. The crowded-winner test, not the thesis, is what is being priced today. | ||||||
| Talen Energy (NASDAQ: TLN) | $361.01 | Wk 15 | ~$399.41 | +10.6% | Nuclear and gas baseload power for data centres; AWS power-purchase optionality | Thesis score 25 Jul. |
| Closed its ECP PJM gas-asset purchase on 15 June. The baseload-for-AI thesis keeps compounding; the next catalyst is the contracted-power optionality being repriced. | ||||||
| Arista Networks (NYSE: ANET) | $147.00 | Wk 20 | $165.45 | +12.6% | EOS software lock-in as AI clusters shift to standard Ethernet; networking at a software margin | 4-wk CLOSED 26 Jun ✓; thesis 28 Aug. |
| Tactical window closed a winner, up 12.6 percent. The real question, whether the networking-layer thesis is right, is still eight weeks out. | ||||||
| USA Rare Earth (NASDAQ: USAR) | $21.00 | Wk 18 | ~$21.00 | ~flat | Only US firm building a complete mine-to-magnet rare-earth chain; dual federal backing | 4-wk scored 13 Jun ✓; thesis 16 Aug. |
| The week China tried to break the thesis and hardened it instead: the 22 June export-control listing restricts imports, not sales. CADE’s Serra Verde ruling, still pending, is the next thing that decides it. | ||||||
| Freeport-McMoRan (NYSE: FCX) | $65.49 | Wk 14 | ~$62.20 | -5.0% | Largest listed copper pure-play on AI build-out demand; ~31% structural supply deficit versus 2035 | Thesis score 18 Jul. |
| Tactically red, structurally intact, the copper deficit did not get smaller because the price did. First thesis verdict lands in three weeks. | ||||||
| MP Materials (NYSE: MP) | $67.21 | Wk 17 | ~$58.00 | -13.7% | US rare-earth mine with a Defense Department cost-plus price floor; strategic-utility re-rating | Thesis score 9 Aug. |
| The lone double-digit laggard, at roughly minus 14 percent after China’s 22 June listing. The curbs restrict its imports, not its sales, so the structural thesis hardens even as the price wears the headline. | ||||||
| Venture Global LNG (NYSE: VG) | $13.08 | Wk 16 | ~$12.43 | -5.0% | Structural Qatar LNG gap independent of Hormuz diplomacy; highest spot exposure among US exporters | Thesis score 2 Aug. |
| Down about 5 percent as sub-80 oil drags LNG-adjacent sentiment, though the real price driver is the Asian gas benchmark, not crude. The structural Qatar-gap thesis is intact; the tape is testing patience. | ||||||
| YPF Sociedad Anónima (NYSE: YPF) | $50.31 | Wk 23 | ~$45.62 | -9.3% | Vaca Muerta shale + ~$20bn Argentina LNG + sovereign re-rating, mispriced as a spot-oil EM cyclical | 4-wk 17 Jul; thesis ~Jun 2027. |
| The four-week driver, oil beta plus Argentine country risk, is still adverse, exactly as declared at entry. The shale-and-sovereign story has not had its moment yet. | ||||||
This week: a scoring week with no new On the Radar name; Arista’s four-week tactical call closed a winner on 26 June, and all eight names are tracked here. The book reads six of eight in the money; MP is the lone double-digit laggard, its Defense Department floor the reason the thesis holds through the drawdown. The first thesis-horizon verdicts, on Bloom Energy and Freeport, fall on 18 July.
| Indicator | Latest | Prior | Direction |
|---|---|---|---|
| Headline CPI YoY (May 2026) | 4.2% | 3.8% | Re-accelerated; energy +23.5% the driver |
| Core CPI YoY (May 2026) | 2.9% | 2.8% | Cooled (0.2% MoM); above target. The 4.2% headline is energy |
| Real Avg Hourly Earnings (12-mo, YoY) | -0.7% | ~0% | Real wages negative over the year |
| Nonfarm Payrolls (May 2026) | +172k | ~+80k cons. | Labour market intact |
| Unemployment Rate (May 2026) | 4.3% | 4.3% | Steady, low |
| ISM Manufacturing New Orders (May) | 56.8 | 54.1 | Fifth month of expansion |
| Retail Sales MoM (May 2026) | +0.9% | +0.4% | Beat; consumer still spending |
| Fed Funds Rate (current) | 3.50-3.75% | 3.50-3.75% | Held 17 Jun; dot plot hawkish, median dot 3.8% |
The long end eased as oil fell and the hawkish-Fed pressure bled off, leaving the curve re-steepened at plus 31 basis points.
| Tenor | Yield | WoW Change |
|---|---|---|
| 2-Year Treasury | 4.09% | -13bps as oil fell and Fed pressure eased |
| 5-Year Treasury | 4.20% | -25bps |
| 10-Year Treasury | 4.40% | -6bps |
| 30-Year Treasury | 4.86% | -19bps |
| 2Y-10Y Spread | +31bps | Positively sloped, re-steepened, the standing red |
| HY OAS | ~278bps | Tight; no distress priced |
| IG OAS | ~76bps | Stable |
Risk came out of the whole complex this week, oil down near pre-war levels, the precious metals and copper all lower, silver hit hardest of all.
| Commodity | Close (18 Jun) | WoW % | YTD % |
|---|---|---|---|
| WTI Crude Oil | $69.23/bbl | -9.6% | +9.5% |
| Gold | $4,078.70/oz | -3.4% | -6.0% |
| Silver | $59.22/oz | -10.6% | -16.1% |
| Copper | $6.14/lb | -3.6% | +8.1% |
| Natural Gas (Henry Hub) | $3.231/MMBtu | -0.1% | -8.1% |
| Baltic Dry Index | 2,524 | -5.1% | +34.1% |
Note: verified Friday 26 June closes. WTI sits near pre-war levels around $69 before Friday’s Hormuz drone strike put a fresh bid under it. Baltic Dry from Trading Economics (26 Jun, 2,524).
| Date | Event | Relevance |
|---|---|---|
| Re-scheduled | Iran nuclear talks (postponed 19 Jun) | WTI sub-$80 confirm or upside-shock risk if it slips again |
| 26 Jun 2026 | ANET 4-week tactical score date | Next On the Radar accountability mark |
| ~late Jun 2026 | CADE ruling on USAR Serra Verde acquisition | USAR thesis-breaker, live risk |
| 30 Jun 2026 | Conference Board Consumer Confidence | A10 consumer-health input |
| 17 Jul 2026 | YPF 4-week tactical score date | New On the Radar call |
| 18 Jul 2026 | BE & FCX thesis-horizon scores | Final dual-horizon verdicts |
| 25 Jul 2026 | TLN thesis-horizon score | Final verdict on the call |
| Mid-Aug 2026 | US-China tariff pause expires; USAR thesis score 16 Aug | Copper, EM, global trade |
| Pair | Rate | YTD % | Driver |
|---|---|---|---|
| USD/TRY | 46.55 | +31.49% | Lira weak; domestic inflation, EM credit pressure |
| USD/ZAR | 16.50 | -5.97% | Rand firm on metals exports, YTD shown against the locked baseline |
| DXY (US Dollar Index) | ~99 | ~flat | Steady as oil and yields eased |
| EUR/USD | ~1.10 | +~2% | Steady as the dollar holds |
| Indicator | Level | Signal |
|---|---|---|
| VIX | 18.89 | Rose 15% on the tech rout; sub-20 but no longer calm |
| MOVE Index (bond volatility) | ~71 | Low; well below 110 caution line |
| HY Credit Spread (OAS) | ~278bps | Below 350bps danger zone |
| S&P % above 200dma | 60.0% | Right at the green/amber boundary |
| 2Y-10Y Yield Spread | +31bps | Positively sloped, re-steepened, standing red |
| Insider Clusters (net selling) | 0 sectors | No net-selling cluster |
| Crash Probability Score | 27.5/100 | No Credible Crash Signal; upper edge of the benign band |
| Company/Event | Data Point | Relevance |
|---|---|---|
| Micron (fiscal Q3) | Revenue more than tripled YoY; gross margin >81%; HBM sold out under multi-year contracts | AI-memory demand and pricing power, confirmed in hard numbers |
| OpenAI listing | IPO reportedly delayed (toward 2027) | Crystallised the AI-cost repricing; supply wave stalls |
| Hyperscaler capex | ~98% of operating cash flow on 2026 consensus; data-centre debt ~$182bn (GS ISG) | Telecom-2000 rhyme; the compute-cost ceiling (Augmenter Pillar 4) |
| Worker vs C-suite (WSJ/GS ISG) | 40% of non-management say AI saves no time vs 2% of C-suite, 38-point gap | Adoption not yet outrunning the S-curve where it matters |
| China rare-earth controls | MP Materials & USA Rare Earth added to export-control list (22 Jun) | Hardens the domestic-supply thesis |
| Flashpoint | Status | WMP Assessment |
|---|---|---|
| US-Iran / Hormuz | Ceasefire breached: four Iranian drones at the Strait on 26 Jun, one striking a cargo ship. Geneva talks collapsed 18-19 Jun; MoU never signed | Signature, not rhetoric. The energy tail was dormant, not dead; watch force-majeure clauses. |
| Qatar LNG / Ras Laffan | 12.8 mtpa offline; 3-5yr repair | Structural supply gap unchanged, independent of any deal. |
| US-China Tariffs | Pause expires mid-August | Swing factor for copper and EM. Watch for extension. |
| Global Liquidity Cycle | Peaked Q4 2025 / early 2026 (Howell) | Qualifies every green credit reading; rollovers absorb liquidity 2026-28. |
Six monthly indicators of the American consumer, updated as new releases drop. No new monthly releases this week; values carry forward. No high-priority flags: confidence above 85, savings above 2.5%, retail sales positive, auto sales above 15.0M. The PCE/savings print and Conference Board confidence both land around 30 June.
| Indicator | Current | Prior | Direction | Release |
|---|---|---|---|---|
| Conference Board Consumer Confidence | 93.1 | Above the 85 flag line | May 2026 (next 30 Jun) | |
| NY Fed 1-Year Inflation Expectations | 3.5% | Elevated, above target | May 2026 | |
| NY Fed % Worse Off Than a Year Ago | 48.0% | Near half of households | May 2026 | |
| Retail Sales MoM | +0.9% | +0.4% | Beat; firm consumer | May 2026 |
| Auto Sales SAAR | 16.2M | Above the 15.0M flag line | May 2026 | |
| Personal Savings Rate | 2.6% | Thin buffer, just above 2.5% flag | Apr 2026 (next ~30 Jun) |
Sources: Conference Board; NY Fed Survey of Consumer Expectations; US Census Bureau; Cox Automotive / JD Power; BEA. The firm retail-sales print sits atop a low savings rate, spending is holding, but the cushion behind it is thin.
Scoreboard closes: verified Friday 26 June 2026 closes from the Supabase wmp_price_daily table (yfinance, confidence verified/api), supplemented by named-source web verification for S&P 500, AGG, TLT and HYG where the automated feed had a same-day gap. Nasdaq 100, LQD and MSCI EM are carried with a flag where neither the feed nor a named source gave a clean 26 June close, and are re-verified on the Monday scoring run. Baltic Dry uses the Trading Economics close (26 Jun 2026, 2,524); MSCI EM is derived from the EEM ETF close times the locked ratio (28.367). All baselines locked 1 January 2026; see the exception report for the carried rows.
Market Probability Dashboard signals: high-yield OAS (~278bps, FRED) and VIX (18.89) from the Supabase macro record; MOVE (~71) carried; ISM new orders from the ISM May report (56.8); percentage of the S&P above its 200-day average (60.0%); insider clusters from OpenInsider (zero sectors). The composite read 27.5 of 100 this week, up from 21.3, at the upper edge of the benign band; the rise is entirely the VIX, with no rubric change.
The oil-shock duration analysis, the hyperscaler-capex and worker-versus-C-suite figures referenced in the Analytical Takeaway and Bubble & Risk Scan are recreated from Goldman Sachs Investment Strategy Group, “US Resilience Resilient,” June 2026; charts not reproduced as images. Practitioner observations are anonymised by role only. Analytical calls are logged at entry and scored at both a four-week tactical mark and a thesis horizon in the Supabase intelligence database (project cxldftvilyhhxcuynhfs). On the Radar entries are framed as “what I am watching and why,” not investment recommendations. Full disclaimer in the footer.