The Rear-View Mirror.

Executive Summary

TL;DR

• Crude rose 15.52 percent to 82.49 dollars, the sharpest weekly rise of its year. The wind-down of the revoked Iran oil waivers expired on Friday, closing the last legal channel for Iranian crude, and the barrel has gone from below 70 dollars in early July to above 82. Petrol follows crude with a lag of two to three weeks, so the July inflation number will carry this move inside it.
• The June inflation print was benign, and it was measured before the oil moved. CPI (the Consumer Price Index, the government’s main measure of inflation) fell 0.4 percent in June, with the yearly rate down to 3.5 percent from 4.2 and core prices, which strip out food and energy, flat on the month. Markets read it as the door to a September cut. The catch: the survey was taken in mid-June, during the ceasefire lull, before the Strait re-closed.
• Japan paid first. The Nikkei fell 6.44 percent, the heaviest weekly fall of the twenty-five assets we track, and emerging markets fell 5.40 percent, while London’s energy-heavy FTSE 100 rose 0.98 percent. Japan imports nearly all of its energy; Britain’s index is full of the companies that sell it. The selloff sorted markets by their petrol bill, which tells you what it was really about.
• Last week’s tell fired against us, and the model beat the narrative. We said a flat core print would reopen the cut path; core came in at zero and it did. Our written lean toward a hawkish tilt was wrong, the second time in two months, while our internal rate model printed Hold both times. The scoring is at the top of the Analytical Takeaway, and the model now anchors the rate view.
• The crash gauge holds at 15 of 100, and this week the calm needs a health warning. The gauge has no energy input at all, and three of its seven signals are carried from earlier dates. It is reading its room accurately; the room it reads does not contain the barrel.
• The call ledger publishes its first two final verdicts, and the loss leads. Freeport-McMoRan closed with the thesis right and the vehicle wrong: copper edged higher over the full holding period while the stock fell 10.86 percent. Bloom Energy closed correct but level with the index, a dead heat. Across every marked call, the book’s cumulative excess return over its benchmarks now stands at minus 0.1 percent. The full ledger, losers first, is in On the Radar.
• Two debuts: rugby and Zipline. The Trophy Asset section runs for the first time, on the 90 million pounds ITV paid for a rugby competition that had never been played. The Case Study is Zipline, the autonomous delivery company valued at 7.6 billion dollars in January, which earned its moat flying blood across Rwanda.

Markets in brief. The S&P 500 closed Friday 17 July at 7,457.69, down about 1.6 percent on the week but still up 8.94 percent for the year. The 10-year Treasury yield, which sets the tone for mortgages and business loans, ended at 4.55 percent, barely changed on the week, a calm that is itself notable in an oil shock. WTI crude finished at 82.49 dollars, up 30.52 percent for the year and now third on our Scoreboard. Gold eased to about 4,013 dollars. The VIX, Wall Street’s fear gauge, rose from 15.03 last week to 18.77, stirring, but below its caution line at 20.

Analytical Takeaway

The gauge reads a calm 15, and that number is precisely the blind spot: it has no energy input, and the one thing that moved this week was energy.

Hold remains the base case, softened toward a cut; the oil shock is the tail that could take the cut away, and the labour market is the tail that could force it anyway.

Reconciliation: after the June print, market pricing swung toward a September cut. Our model reads Hold with a bias toward a cut, and it refuses to lurch: only one of its three blocks moved this week, and its own discipline requires two to agree before the view changes. We treat the market’s cut pricing as sentiment to be tested, not as an input to follow.

How the 15 is built: each signal scores 0 when green, 5 when amber and 10 when red; the scores are weighted and summed. The weights, the thresholds, this week’s full working and the three inputs that are carried rather than fresh are printed in the appendix, A6, so you can recompute the number yourself.

Last week’s tell, scored first, and it went against us. The tell said a core reading of 0.3 percent or above would confirm our hawkish tilt, while 0.2 or below would reopen the cut path. Core came in flat, zero on the month, below the threshold with room to spare, and market pricing swung from two-thirds odds of a September hike to a live September cut. Both legs fired dovish. The Hold was right; the hawkish tilt we attached to it was wrong, and this is the second time in two months the written rate view has leaned hawkish into a benign inflation print. The genuinely useful detail is who did not make the mistake: our internal rate model printed Hold, no tilt, both times. The tilt was editorial gloss layered on a model that was not asking for it. The model was built after the June error precisely to prevent this. This week it worked, and the narrative around it did not. From this edition the model anchors the rate view, and any departure from its reading has to be argued in print, not assumed.

The rear-view mirror

On Monday 13 July, WTI crude jumped 9.23 percent in a single session, after the revoked Iran waivers and fresh attacks on tankers re-closed the Strait of Hormuz in practice. On Tuesday morning the June CPI landed: prices down 0.4 percent on the month, energy down 5.7 percent, the yearly rate at 3.5 from 4.2. The market chose the Tuesday number over the Monday one. Bond traders priced a September cut, and the AI-power complex, the year’s most crowded trade, sold off hard into it.

The problem is arithmetic, not opinion. The June price survey was collected in mid-June, inside the ceasefire lull, with crude below 70 dollars. The waivers were revoked on 7 July; the barrel closed this Friday at 82.49. Trading on June’s inflation number today is like navigating with a photograph of the road taken last month: the picture is accurate, and the road has moved. The 5.7 percent fall in energy prices that made June look so cool is already running in reverse, and petrol follows crude within two to three weeks. The July CPI, due 11 August, is the print that will carry this week inside it.

Japan is the cleanest way to see what the market actually repriced. The Nikkei fell 6.44 percent, the heaviest fall of the twenty-five assets we track, and Japan imports nearly all of its energy: every barrel it burns arrives on a ship, priced in dollars. Emerging markets, on balance also importers, fell 5.40 percent. Britain’s FTSE 100, stacked with the majors that sell energy, rose 0.98 percent, and Switzerland’s defensive index gained too. Markets did not sell risk in general this week; they sorted the world by its energy bill. And the sorting reaches the household directly. A mortgage path is being priced today off a June inflation reading while the July petrol bill climbs, the same rear-view trade in miniature. If the cut now priced rests on a number the oil market has already overtaken, the borrower waiting on that cut is the last to find out.

The rate read, governed by the model. The model’s dial moved from 9.0 to 4.0 this week, further inside the Hold band, with the bias tilting toward a cut. Exactly one thing moved it: the soft June services-inflation reading. That matters for what we will not do. The model’s rule requires two of its three blocks, inflation, labour and markets, to agree before the view itself changes, and only one block moved, so the view stays Hold, softer in tone, without lurching dovish; one block moving is noise. Two supporting facts, stated plainly. The panel behind the read is fully scored for the first time, no provisional cells. And the Taylor Rule benchmark it anchors on (a mechanical formula for where the policy rate should sit given inflation and unemployment) still prescribes a rate about 1.33 percentage points above the funds rate, a stable gap. A cut is not yet rule-justified, however the market prices it.

The competing explanation, named. We have told this week’s selloff as an oil story, and we should say what else it could be: positioning. The AI-power complex was the most crowded trade in the market, and crowded trades unwind on their own timetable. TSMC grew net income 77 percent and added a further 100 billion dollars to its American investment programme this very week, taking the Arizona commitment to roughly 265 billion dollars, and the complex sold off anyway, which is what profit-taking looks like. The tell that separates the two readings: an oil-driven selloff sorts markets by energy exposure, a positioning unwind sorts them by crowding. This week the sort ran by energy, Japan and the importers fell hardest while the FTSE rose. If next week the energy-heavy, never-crowded indices give back their gains while oil holds, the positioning read wins, and we will say so.

Where I could be wrong (the standing self-check): First, the model we have just made the anchor is fed by the same rear-view data this edition warns about: its inflation cells are June-vintage. If the July prints re-accelerate hard, the model will be late by construction. Its own falsification line names the exit: a hot core reading on a three-month view, or the 2-year above 4.30 percent, sends the read back toward hawkish. Second, the calm we report may simply be mismeasured: the crash gauge has no energy input and three of its seven signals are carried, so this week’s specific risk is largely invisible to it. The tell: high-yield spreads widening through 350 basis points, from 271, while the gauge still reads green.

Zipline: The First Ascent

The founder, in one archetype: the Jungle Explorer. The pioneer who hacks a new category out of ground nobody else will enter, until the trail he cut becomes the road everyone else drives. With Keller Rinaudo Cliffton it is nearly literal: he built autonomous logistics in rural Rwanda, where no incumbent operated and no regulator had a rulebook, and only then brought it home to American suburbs.

The problem, first. In a Tanzanian health institute in 2014, Rinaudo was shown a database of preventable deaths: people who died not for want of a cure but for want of a delivery. The blood, the vaccine, the antivenom existed, just not where and when it was needed. A graduate student had built a text-message system so rural health workers could request supplies; thousands of requests had come in, and there was no way to fulfil them. The problem was never medicine. It was logistics.

Who he is. Rinaudo was not a logistics man or an aviation man. He was a former professional rock climber with a Harvard science degree who had co-founded Romotive, maker of a smartphone toy robot, backed through Techstars Seattle and a funding round led by Sequoia. He lost faith that toy robots mattered, and in 2014 he did the thing founders are told never to do: he pivoted a funded company off its product entirely, onto autonomous medical delivery in a country with sparse roads and no drone rulebook. In climbing, a first ascent means finding a line up a face everyone else called unclimbable; once cut, it becomes a named route others follow. That is the operating style here. Rwanda signed in 2016, and Zipline’s fixed-wing aircraft now carry three-quarters of the blood supply used outside the capital, Kigali, with a nationwide expansion signed in February. Then Ghana, the United States, Nigeria, Japan, Kenya and Côte d’Ivoire. The company says it has flown more than 135 million autonomous miles and made more than 2.5 million deliveries, roughly one every twenty seconds by now, with about seventy percent of flights now American. In January it raised at a valuation of 7.6 billion dollars, in a round led by Valor Equity Partners.

The moment of real uncertainty. It is on the record, and it is recent. Ghana, a Zipline country since 2019, ran up roughly 22 million dollars in unpaid bills as its currency slid and a new government argued most of the flights were not essential; in December 2025 Zipline shut half of its Ghanaian distribution centres. That is the sharpest lesson in this story: when your customer is a single government and its currency, your mission is only as durable as its balance sheet. And the current bet is heavier. The next chapter is staked on Platform 2, a hovering aircraft that lowers packages to suburban doorsteps on a tether, live with Walmart in Dallas-Fort Worth since last year and launching with the Cleveland Clinic this month. Doorstep delivery is a far harder economic problem than life-or-death rural logistics, the competition is Alphabet’s Wing and Amazon’s balance sheet, and scale in America depends on a federal drone-traffic rule, known as Part 108, that keeps slipping.

The valuation, kept separate from the admiration. The franchise is genuinely impressive; the price is a different question. The 7.6 billion dollar mark rests on undisclosed economics: Zipline does not publish its revenue, its losses or its cost per delivery, and on rough delivery-count arithmetic the revenue is plausibly modest against that valuation. Our own analysis framework scores it 6.9 out of 10, a watchlist verdict: a top-decile company by impact and operational scale that is not obviously a top-decile entry at this price, because the mark pre-pays for American regulation and doorstep economics both breaking right. Conviction in a company and conviction in its price are two separate tests. This profile passes the first and withholds the second.

What it teaches. Two things, both transferable. First, the category was created in the place nobody else would go: the operational data and regulatory trust now protecting Zipline in America were earned in Rwanda, where its competitors were not. The hardest, least glamorous market was not the detour; it was the moat. Second, impact and durability are different variables: the healthcare mission is real and measured, and Ghana shows it does not by itself make a durable business. Admire the achievement. Keep a clear eye on the price.

Our screening framework scores Zipline 6.9 out of 10: a watchlist, not a buy. A note on the numbers: valuation, delivery and mileage figures are from the company’s July release and round disclosures; impact and safety statistics are company-attested, not independently audited; revenue, burn and unit economics are undisclosed.

The Week That Was

The barrel wrote the week. Crude defined the week's trading. WTI capped a fifteen percent weekly rally, most of it in a single Monday session, to close at 82.49 dollars, pushing its gain for the year past 30 percent and up to third on our Scoreboard. The trigger was legal, not military: the wind-down of the revoked Iran oil waivers expired on Friday, closing the last sanctioned channel for Iranian crude after attacks on three tankers ended the reopening experiment. Natural gas barely moved, down about 1 percent on the week, a reminder that this is an oil-specific squeeze rather than a general energy panic.

Equities sorted themselves by energy bill. Japan’s Nikkei fell 6.44 percent and emerging markets fell 5.40 percent, the two heaviest falls on the board, and both are energy importers. The Nasdaq 100 fell 4.13 percent as the AI-power complex kept repricing, and the S&P 500 slipped 1.55 percent. On the other side of the sort: Hong Kong rose 1.60 percent, London’s FTSE 100, heavy with oil majors, added 0.98 percent, Switzerland gained 0.76 and India edged up 0.53. Small caps were nearly flat, down 0.52.

The quiet dial was the bond market. An oil shock is textbook inflation news, yet the 10-year Treasury ended at 4.55 percent, up a single basis point on the week, and long-bond funds were flat. Read strictly, the bond market treated the oil move as a tax on growth at least as much as a push on prices. That restraint is holding the equity market up: the day the long end starts trading oil as inflation is the day this selloff changes character. Gold slipped to about 4,013 dollars and silver fell 6.31 percent, an odd pair of moves for an inflation-scare week, and consistent with the growth-tax reading.

Earnings said the machine still works. TSMC, the Taiwanese foundry that manufactures most of the world’s advanced AI chips, grew net income 77 percent year on year and added a further 100 billion dollars to its Arizona programme, lifting the total commitment to roughly 265 billion dollars. The big American banks broadly beat expectations. June retail sales rose 0.2 percent, and 0.7 percent once falling petrol-station receipts are stripped out. The week’s selling was not about earnings; the earnings were fine. It was about the price of the fuel underneath them.

And the freight pulse cooled. The Baltic Dry Index, the cost of moving raw materials by sea and a real-economy pulse-check, fell 6.52 percent from last week’s spike, still the year’s best line at plus 46.23 percent. Crypto sat the week out: Bitcoin was flat, down 0.36 percent, and Ethereum rose 2.52.

Bubble & Risk Scan

The dashboard is green on every dial it reads, and the week’s defining move, a fifteen percent rise in the price of oil, happened on a dial it does not have.

The standing qualifier, sharper than usual this week. Every reading above is honest, and the instrument itself has a blind spot you should hold onto. The gauge has no energy-price input, no long-end yield input and no credit-concentration input, and three of its seven signals, bond volatility, factory orders and insider selling, are carried from earlier dates. It is a smoke alarm wired to the kitchen while this week’s risk is the petrol in the garage. The alarm being silent is information; it is just not information about the garage.

The Speed of Now

The number is 77 percent. TSMC, the Taiwanese factory that makes most of the world’s advanced AI chips, reported on Thursday that its net income grew 77 percent in a year. It also added a further 100 billion dollars to its Arizona investment, taking the total programme to roughly 265 billion dollars. That is the speed of the build-out stated by the one company positioned to know: demand for AI silicon is still running ahead of the machine that makes it.

The counter-number is a budget cap. The more interesting speed signal ran the other way this week: Uber and Meta, two of the heaviest corporate users of AI, have reportedly begun capping what their own teams may spend on AI model usage. Adoption so fast it needed a budget is still adoption, but a metered utility grows differently from an unmetered one, because the cost ceiling arrives before the capability ceiling. Hold both facts at once and you have this year’s AI economy in miniature: the factory cannot keep up, and the customers are watching the meter.

Geopolitical Watch

The week’s biggest geopolitical number was written by an agency, not an army: 6.5 trillion dollars. On Wednesday the International Energy Agency published its assessment of China’s rare-earth export curbs. If fully implemented, the controls would put roughly 6.5 trillion dollars of annual production outside China at risk, with car manufacturing the largest single exposure at more than 3 trillion (Bloomberg, Semafor, 16 July). A further 300 billion or so is tied to planned graphite controls. Be precise about what this is, because the classification is the analysis: it is an intergovernmental agency quantifying controls China announced last October and then delayed by a year. It is not a new Chinese measure, and it is not an escalation. Nothing changed in Beijing this week. What changed is that the West read its own exposure aloud, with a number attached, and a named exposure gets priced where an abstract one gets ignored. The mechanism for a reader: rare earths are seventeen metals used in tiny quantities that sit inside almost every motor, sensor and magnet. A supply threat therefore prices not the metals but everything downstream of them, which is how a niche mining story becomes a 6.5 trillion dollar one.

It lands directly on this book. Our two rare-earth calls, MP Materials and USA Rare Earth, were both added to Beijing’s export-control list on 22 June, and both fell double digits again this week as the sector repriced. The IEA report did not change either company’s position by a single tonne; it changed how visibly the chokepoint is priced. Both calls are marked, loudly, in Portfolio Watch below.

Hormuz, second this week, and re-verified at the source. The sequence, from the sanctions record rather than from memory: in late June Washington issued a general licence allowing limited Iranian oil transactions under the June memorandum. After Iranian attacks on three tankers in the Strait, it revoked that licence on 7 July, replacing it with a wind-down authorisation that expired on Friday 17 July, the day of this edition’s closes. As of Friday, the legal channel for Iranian crude is shut again, which is the single largest reason the barrel ended the week above 82 dollars. The June agreement survives on paper; its oil provisions do not survive in practice. The spring taught us that a ceasefire settles nothing by itself. The sequel is now on the record too: a signature is not compliance. The residual risks remain the slow ones, missile and production reconstitution, and the leading indicator to watch is still force-majeure activations in shipping insurance, not the headlines.

The cross-check that keeps both stories honest. A genuine supply catastrophe would show up in the long end of the bond market and in credit spreads. Neither moved: the 30-year rose modestly to 5.09 percent and high-yield spreads sat still at 271 basis points. The market is treating both chokepoints, the Strait and the rare-earth list, as priced frictions rather than ruptures. That is also exactly what the pricing would look like the week before it was wrong, which is why the insurance clauses, not the indices, remain the tell.

Rugby Sells Its Structure

In January, ITV agreed to pay 90 million pounds for the exclusive British rights to a rugby competition that had never been played. The inaugural Nations Championship kicked off on 4 July: twelve national teams, a real table, home-and-away windows in July and November, and finals at Twickenham at the end of November. It is run as a joint venture between the Six Nations and the southern-hemisphere unions, and played every two years. For a century, international rugby outside the World Cup was an unstructured calendar of one-off matches, no standings, no cumulative consequence, unsellable as a season-long product. ITV did not buy rugby. Rugby was always there. It bought a structure.

That is this section’s thesis in its purest form. A scarce, prestige asset becomes investable the moment a structure arrives that capital can underwrite. Collectable cards traded for decades in a market nobody could price, until professional grading turned an opaque, peer-to-peer hobby into a verifiable, tradeable one; the institutional bid followed the grading, not the cards. International rugby fixtures are a fixed-supply asset, a national team can only host so many afternoons a year, and the Nations Championship is the grading service: it converts one-off friendlies into a scheduled, rights-bearing season. The bid moved instantly, and revealingly. TNT Sports, the pay-TV incumbent, held first refusal and declined, reportedly because a biennial product delivers inventory only every other year; ITV took it free-to-air. The audience case was already on the record: ITV’s England and France Six Nations broadcast peaked at six million viewers against roughly one million for England and New Zealand behind TNT’s paywall. Reach, not subscription, is what a young competition needs first.

Two live ironies are playing out as the first season runs. The first is Qatari. Qatar’s government offered guaranteed returns of 800 million pounds to host the first four finals; the unions turned it down, citing low expected attendance and Qatar’s human-rights record, and took Twickenham instead, though Qatar still hosts in 2028. Meanwhile Qatar Airways’ title sponsorship of the competition, 80 million pounds over eight years and agreed in outline as early as January 2025, was still unsigned at kick-off and is paused. Qatari figures reportedly judge it inappropriate to announce amid the regional fallout from the American strikes on Iran. Follow that thread and it lands in this edition’s lead story: the competition sold its structure, and the single biggest cheque on its table is hostage to the same conflict that moved the oil price fifteen percent this week. The second irony is corporate. On 6 July, Sky, owned by Comcast, agreed to acquire ITV’s media and entertainment business for up to 1.6 billion pounds. The free-to-air broadcaster that bought these rights precisely to take rugby out from behind a paywall is itself being absorbed by a pay-TV operator. We searched for reporting on what that means for the rugby rights and found none. That is reported here as not found, which is not the same as no impact.

Now the bear case, at full volume, because this section does not write fan letters. English club rugby, the layer beneath the internationals, is a financial ruin. Three of the thirteen clubs that started the 2022-23 Premiership season went bust: Worcester and Wasps in 2022, London Irish in 2023. Wasps carried roughly 95 million pounds of debt, and its only profitable modern year was the year the private-equity investment itself arrived. Cumulative Premiership club losses exceed 500 million pounds over twenty-five years. The remedy chosen to make the league underwritable was to abolish promotion and relegation, and that deserves to be said plainly: the fix for a competition that could not fund itself was to remove sporting jeopardy, the very thing the product sells. Nor has professional money obviously beaten the game. CVC, the private-equity firm, put roughly 700 million pounds into club and international rugby across three deals, including 365 million pounds for exactly one-seventh of the Six Nations. Its attempt to sell a stake in the sports vehicle holding those positions at a 9 billion euro valuation failed, with would-be buyers reportedly citing the rugby and French football holdings. It settled for a roughly 7 billion euro refinancing instead. And above all of it hangs a courtroom: 295 former players are suing the game’s governing bodies in the London High Court over dementia and other brain injuries. The governing body’s defence, filed in February, denies liability and calls the injuries a foreseeable and inherent risk of the sport. No damages figure is public, and we will not invent one. It is an unquantified contingent liability attached to the same balance sheets whose rights are being sold.

The sharpest number in the file explains the whole trade. England’s rugby union earned 228 million pounds in the 2024-25 season from seven home internationals, while the entire commercial business of the Springboks, the world champions, was implicitly valued at 375 million dollars in the private-equity offer their union rejected in December 2024. Read together: seven London afternoons earn more in one season than the market then priced the best team on earth at. Home-match inventory is the asset, and the Nations Championship is a re-bundling of exactly that inventory into something a broadcaster can underwrite. One calibration, labelled honestly as our own arithmetic and not a published valuation: extrapolating CVC’s one-seventh stake would mark the whole Six Nations near 2.55 billion pounds, before any control discount. The honest conclusion runs both ways. The structure is real and the rights market has already voted; the operating layer beneath it keeps setting money on fire, and a scarce asset run by broke operators is the oldest trap in the trophy-asset category.

The garnish, dated for the record: the competition’s third round plays this weekend, straight into the football World Cup final at MetLife Stadium on Sunday, both after this edition’s Friday close. A new product choosing the most contested attention weekend of the northern summer is either confidence or a scheduling accident, and the audience figures, not the match reports, will tell us which.

The Light That Stayed On

In 2000, roughly 1.4 billion people, close to one person in four then alive, had no electricity at all. By 2022 that number had fallen to about 750 million, fewer than one in ten, and it fell while the world added two billion people. More than nine in ten human beings now live with the light on. It is one of the great infrastructure achievements of the age, and it happened so gradually that no single week’s news ever carried it.

This week the market repriced what energy costs. Hold that against the century’s quieter energy story, which is who has it at all. India connected essentially all of its villages and hundreds of millions of its citizens inside a generation. Bangladesh, Indonesia, Kenya and Ghana each moved from partial coverage toward near-universal access. The remaining darkness is concentrated and honest to name: most of the roughly 750 million people still without power live in Sub-Saharan Africa, and progress there has slowed since 2020 as borrowing costs rose. The chart below carries both the triumph and the unfinished end of it.

The reason this belongs in an oil-shock edition is that the two stories are the same story at different speeds. The price of a barrel this week is set by a strait and a sanctions deadline; the share of humanity with electric light is set by decades of grid-building, collapsing solar costs and state capacity, and that line does not notice ceasefires or waivers. A reader worried, correctly, about what 82 dollar oil does to a July inflation print can hold a second, equally factual thought at the same time. The fraction of our species that cannot refrigerate medicine or light a classroom after dark has nearly halved in a generation. The tape measures the week. This line measures the era, and the era is winning.

While markets reprice the cost of energy, the share of people who have it at all has climbed from around three-quarters in 2000 to more than nine in ten.

The Displacer vs the Augmenter

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.

A clean sweep this week, on quiet Displacer evidence. Four pillars, one point each:

• Pillar 1, adoption speed, the Augmenter. The week’s AI news was demand ramping through existing channels, TSMC’s net income up 77 percent on AI-chip orders, not a capability jump or fresh evidence of AI improving AI. Build-out, not acceleration of the substitution timeline.
• Pillar 2, labour, the Augmenter. No named white-collar layoff wave citing AI as the cause surfaced this week, and the sharpest labour-adjacent signal cut the other way: Uber and Meta reportedly metering their own internal AI usage is budget friction inside the heaviest adopters, not displacement.
• Pillar 3, type of shock, the Augmenter, for the first time in six weeks. The fresh prints were consumption-positive: June retail sales up 0.2 percent, up 0.7 excluding petrol, 6.7 percent higher than a year ago, with the savings rate at 3.0 percent. No new consumption-gap evidence arrived. The honest footnote: the household-strain readings that scored this pillar for the Displacer through June, elevated inflation expectations, concentration at the index top, are carried rather than reversed, and the July petrol bill will test this pillar hard next month.
• Pillar 4, compute cost, the Augmenter. TSMC adding a further 100 billion dollars to its Arizona programme with the 30-year Treasury at 5.09 percent is the binding-ceiling thesis in one sentence: the build-out is financed at elevated long rates, and the chipmaker’s own record margins show compute prices rising, not falling toward the point where a machine undercuts a salary.

Week 27: the Augmenter 4, the Displacer 0. Running total, the Augmenter 25, the Displacer 7. A sweep records an evidence-quiet week for the Displacer, not a settled argument; the last sweep, in May, was followed within a fortnight by the Displacer’s first genuine points. What would flip the score: a named white-collar layoff round citing AI as the structural cause hands the Displacer Pillar 2 next week; a fresh consumption-gap print, real retail spending turning negative while mega-cap margins expand, takes Pillar 3 straight back; and a falling long rate that cheapens the build-out’s financing would weaken the Augmenter’s hold on the compute ceiling.

On the Radar

What I am watching and why, not a recommendation to buy or sell. No new name this week: the one candidate the screen surfaced had not completed our mandatory pre-publication analysis by press time, and a name that has not been through the full framework does not run, whatever its chart looks like. This week the section does its other job: the first two final verdicts on the book, and the ledger in full.

The ledger first, led by the number that hurts. Across every call this publication has made and marked, the cumulative excess return over benchmark, what the calls earned beyond what you could have had by simply holding the obvious alternative, stands at minus 0.1 percent. Twenty-one calls in, the book has, in aggregate, beaten nothing it could have held instead. The rest of the record: thirteen of twenty four-week timing checkpoints correct, 65 percent, counting all twenty tactical checkpoints logged in the call record since Week 14. Both thesis-horizon verdicts to date closed correct, two of two, and there are no double-misses. On the five-point verdict scale the book stands at none vindicated, two working, one open, one failed-soft, none failed-hard. The average win beat its benchmark by 16.1 percentage points; the average loss trailed by 5.4. One exclusion, stated openly: YPF’s four-week checkpoint was scored this week at minus 1.25 percent, a timing miss now counted in the thirteen-of-twenty. Its benchmark comparison could not be honestly computed, because the benchmark price for its entry date is not in our record, and we will not substitute a searched number. YPF therefore sits outside the excess-return figure until that gap is filled, and we say so rather than let the aggregate quietly flatter.

Freeport-McMoRan: thesis right, vehicle wrong, and it leads because it is the loss. In April we said copper was the market’s cleanest structural shortage, and we expressed it through the largest listed copper miner at 65.49 dollars. The final verdict fell this week, on the terms we pre-committed at entry: the invalidation was copper breaking below its floor near 5.80 dollars, and it never fired. Copper finished the holding period at 6.22 dollars, up 1.91 percent. The commodity thesis was correct. The stock closed at 58.38 dollars, down 10.86 percent, trailing the metal it digs by 12.77 percentage points. On the five-point scale that scores a minus one, failed-soft: behind benchmark at its time-stop with no invalidation fired. The lesson is exact and worth keeping: a mining company is a commodity view plus a balance sheet, a cost curve, a jurisdiction and a capital-spending cycle. Every one of those is a way to be right about the thesis and still lose money. If your view is on a commodity, the commodity is usually the cleaner expression than the company that digs it. That rule is now written into our entry discipline: a commodity view wearing an equity costume has to justify the costume at entry.

Freeport-McMoRan (NYSE: FCX) · entry $65.49 (Wk 14) · close $58.38 (17 Jul) · -10.86% vs copper +1.91% · final verdict: thesis held, vehicle failed, scored -1

Bloom Energy: correct, and a dead heat. The book’s founding call, on-site fuel-cell power for AI data centres deployable in months rather than grid-queue years, closed its thesis window this week at 214.96 dollars against a 207.10 entry, up 3.80 percent. The pre-committed invalidation never fired, so the thesis is scored correct, and then the benchmark does its honest work: the S&P 500 returned 4.65 percent over the same window (7,126.06 at entry to 7,457.69), so the call’s excess return is minus 0.86 percent. A dead heat, scored zero. The analytical read held for three months and earned nothing over the index, and both halves of that sentence belong in the record with equal weight. Two loud events sit outside this verdict and cannot reopen it: the short-seller report on Bloom’s scandium supply chain (published 8 July, rebutted by the company on 9 July) and the 28 July earnings. A closed call stays closed; whatever happens on the 28th happens to a company we no longer score.

Bloom Energy (NYSE: BE) · entry $207.10 (Wk 14) · close $214.96 (17 Jul) · +3.80% vs S&P 500 +4.65% · final verdict: thesis correct, excess -0.86%, scored 0

What the founding pair teaches. Both of the book’s first two theses were analytically right, and together they beat nothing: one lost to the metal it tracked, the other drew with the index it was meant to outrun. That is what the excess-return column is for. A scoring system that cannot hurt is marketing, and this month it hurts: the aggregate says that twenty-one calls of genuine analytical effort have so far added a rounding error under zero against simply owning the alternatives. The tactical hit rate says the timing sense is real; the excess column says the vehicle selection has eaten the edge. The fix is not a new framework, it is the rule above, written at entry, every time. The next verdicts land quickly: Talen next Friday, then Venture Global, MP Materials, USA Rare Earth and Arista through August. The ledger will be republished with each one.

And Finally

Sunday will stage a small experiment in the economics of attention. The football World Cup final kicks off at MetLife Stadium, the largest single audience of the sporting calendar, and world rugby’s brand-new competition, having waited a century for a proper structure, has scheduled its third round for the same weekend. Somewhere a rights executive who paid 90 million pounds for exclusivity is discovering that the one thing you cannot buy exclusively is Sunday. Economists call this a positional externality; broadcasters call it the schedule; the rest of us call it two remotes and a household negotiation. The market lesson is the one this section always ends up at: scarcity is only valuable if the scarce thing is attention, and attention is the one asset class that refuses to be securitised. Next week we will know whether rugby’s new table survived contact with football’s oldest one.

A gentle June print stole the show,
The cut trade was ready to go.
But the tape that it read
Was a fortnight-old thread,
And the petrol pump already said no.

Three things to watch next week, all measurable, none resolved. First, the barrel: does WTI hold above 80 dollars into Friday, or give the shock back? Second, the 2-year Treasury at 4.10 percent: the rate market’s referendum on whether the September cut survives. Third, the first core inflation checkpoint, the core PCE reading on 31 July, which tells us whether the cooling that so excited markets survives outside the rear-view mirror. If all three move together, oil high, the 2-year rising, core firming, the disinflation trade of mid-July will be fully unwound and the cut is gone; if all three break the other way, this week was a tremor and the cut lands. If they diverge, oil high while core cools, we get the genuinely hard tape: a Fed watching yesterday’s inflation fall while tomorrow’s builds. Next week’s edition also carries the scheduled half-year check on our Repricing thesis, and Talen’s final verdict. The rear-view mirror has had its week. The windscreen gets its turn on 11 August.

Portfolio Watch, Active Calls

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. Every close below is the verified close we recorded on the date shown beside it. Two names have left the table this week: Bloom Energy and Freeport-McMoRan closed their thesis windows and received their final verdicts in On the Radar above.

This week: seven active calls tracked here after Bloom Energy and Freeport-McMoRan closed at their thesis horizons, both reported in full above. Four of the seven are in the money (Arista, MUFG, VG, Talen) and three are behind, with the losers named as loudly as the winners. MP Materials, down 32.7 percent, is the book’s biggest loss, and USA Rare Earth sits on the same repriced thesis. The marks come quickly now: Talen’s verdict next Friday, MUFG’s first checkpoint 31 July, then Venture Global, MP, USA Rare Earth and Arista through August.