CPI at 3.8 percent. The Taylor Rule waiting. The market has priced the chair. It has not priced the first press conference.
"The price of energy is the price of everything else." — Anthony Rosenthal
TL;DR
Markets in brief. The S&P 500 closed at 7,408.50, up 8.2 percent year-to-date. The Hormuz energy premium is now embedded in equity multiples through the energy sector, but rate sensitivity is pulling in the opposite direction in growth names. The 10-year yield re-inverted at 4.52 percent following the CPI print, with the 2-year at 4.62 percent. Gold at 4,547.89 dollars continues its structural bid from central bank diversification.
The week's hinge variable. Watch the first Warsh Fed statement. Not the rate decision itself — the language around the reaction function. If Warsh signals that 3.8 percent CPI is a ceiling not a floor, the long end will rally and growth equities breathe. If he signals that the Taylor Rule is the new anchor, we reprice everything above 20x earnings.
The Taylor Rule is not a policy tool. It is a mirror. When the gap between where the Taylor Rule says rates should be and where they actually are reaches 125 basis points — which is where we are today, with standard Taylor Rule formulations implying a rate of around 5.5 percent against an actual 4.25 to 4.50 percent — the mirror becomes uncomfortable to look at in public.
Kevin Warsh knows this. He helped design the post-GFC framework that made the Fed more reactive and less mechanical. But his confirmation at 54 to 45 tells you something about the political landscape he is entering. He has no margin for a dovish misstep. The first press conference will be watched with the intensity usually reserved for earnings calls at companies that have not reported in six months.
Three scenarios for the next FOMC cycle:
| Scenario | Probability | Trigger | 2Y Impact | 10Y Impact | Equity Impact |
|---|---|---|---|---|---|
| Warsh Holds | 45% | Core PCE softening in June print | -10 to -20bps | -5 to -15bps | +3 to +5% |
| Warsh Hikes 25bps | 35% | CPI at 3.8% forces Taylor Rule compliance | +30 to +40bps | +15 to +25bps | -6 to -10% |
| CPI Relief Rally | 20% | June CPI surprises to the downside | -25 to -35bps | -20 to -30bps | +5 to +8% |
The base case is a hold with hawkish language — a Powell echo but with a harder edge on the reaction function. What the market is systematically underweighting is the scenario where Warsh moves. Not because moving is the right call in isolation, but because the political optics of a Fed chair confirmed by a nine-vote margin failing to act on a 3.8 percent CPI are worse than the economic cost of a 25-basis-point hike.
Cross-asset divergence. Equities have held because energy companies are printing cash at WTI 105.42 dollars. That mechanical support is masking the rate sensitivity in the rest of the market. The divergence between energy-heavy indices and pure-play growth is now 11 percentage points year-to-date. This is a distributional story: the capital that benefits from WTI 105 is not the same capital that suffers from a 4.52 percent 10-year. Both can be true simultaneously. The error is treating the index as a single view.
2026 Thesis Check-In — Week 18. The Year of the Repricing thesis is tracking. As of today: the S&P 500 is up 8.2 percent, WTI is up 66.8 percent year-to-date, Gold is up 4.8 percent, the Bloomberg Aggregate is down 3.2 percent, and HY-IG spreads have widened 25 basis points from the January baseline. Three of five asset classes are now directionally divergent. The falsifiable claim made in Week 17 has already been met in May. The question is whether the divergence widens or mean-reverts. The energy-inflation loop argues for the former.
CPI 3.8 percent, above consensus. Warsh confirmed 54 to 45. Taylor Rule gap 125 basis points. WTI 105.42 dollars. 10-year re-inverted at 4.52 percent.
Warsh will use hawkish language at the first press conference regardless of the rate decision. The political mandate from a 54-to-45 confirmation is to look credible on inflation, not to ease the growth trade.
A June CPI print below 3.5 percent changes everything. Also: a Hormuz diplomatic resolution that takes WTI below 90 dollars collapses the energy inflation component and makes Warsh's job substantially easier. The diplomatic calendar points to early July as the earliest plausible moment for any partial framework. Full normalisation is a midsummer story at best — meaning the energy inflation variable persists through Q2 and into Q3 regardless of diplomatic tone.
In 2019, a materials scientist named Chad Edwards was staring at a spreadsheet of failed battery cathode experiments when he had the insight that would become the foundation of a platform that major industrial groups would pay eight-figure sums to access. The spreadsheet was not the problem. The problem was that materials science had accumulated roughly 150 years of experimental data across thousands of journals, and no human team could hold all of it in working memory simultaneously. Edwards's insight was that this was not a chemistry problem. It was a retrieval problem.
Edwards left his research role at Cambridge and spent two years building what he called a "materials memory" — a large language model trained not on text but on experimental outcomes. Every paper, every failed synthesis, every molecular interaction log that had ever been digitised and published. The model learned, in effect, to read the periodic table the way a chess engine reads a board position: not by memorising moves but by understanding which configurations lead to interesting states.
The strategic uncertainty. By 2023, CuspAI had a working prototype that could predict novel material properties with roughly 70 percent accuracy before synthesis. The question that nearly ended the company was whether 70 percent was good enough. Industrial customers have zero tolerance for synthesising a material that does not behave as predicted. The first twelve months of customer conversations were a prolonged exercise in the gap between "impressive in a demo" and "useful in a factory."
Edwards's answer was not to improve the model's accuracy. It was to reframe what accuracy meant. The model was not there to replace the lab. It was there to reduce the number of experiments by telling chemists which 30 percent of the search space was definitely not worth running. That reframe changed the commercial conversation entirely.
The raise. In September 2025, CuspAI closed a Series A of 100 million dollars at a 520 million dollar post-money valuation, co-led by NEA and Temasek, with participation from major industrial strategic investors whose materials bottlenecks the platform directly addresses. The strategic logic is straightforward: large industrial groups with decade-long materials development cycles gain access to a platform that compresses that timeline without requiring them to rebuild their internal R&D infrastructure from scratch.
What is transferable. Edwards's insight — that the bottleneck was retrieval, not synthesis — applies to any domain where experimental data has accumulated faster than human capacity to integrate it. Pharmaceuticals. Semiconductor process chemistry. Structural materials for aerospace. The architecture — training on experimental outcomes rather than text descriptions — is replicable across domains where outcomes are measurable and publishable.
CuspAI's platform was designed for industrial chemistry. It is also, structurally, a tool for accelerating the discovery of materials with commercially undesirable properties — explosives precursors, novel energetics, materials with extreme thermal stability that have no obvious civilian application. Edwards has built an export control compliance layer into the platform, and its industrial strategic partners run their own screening. But the dual-use risk is inherent to the capability, not the application. As the model improves, the gap between "accelerating battery development" and "accelerating something else" narrows. The EU's AI Act has no specific provision for materials discovery platforms. The UK's AISI is aware of the category. The regulatory lag is the risk.
Anthony Rosenthal analysis score: 6.5 / 10 — WATCHLIST / AMBER. Strong domain insight and credible corporate validation from major industrial strategic investors. The dual-use risk and regulatory lag are manageable but real. The 520 million dollar Series A valuation is aggressive for a company with no revenue at scale. Watch for the first disclosed commercial contract with a named strategic partner — that is the validation event the market is waiting for.
Monday. Kevin Warsh was sworn in as the 17th Chair of the Federal Reserve. His opening statement contained the line that will be quoted for the next twelve months: "The work is not finished." Powell said something similar in 2022. The difference is that Powell was raising rates to start the work. Warsh is inheriting an environment where the work may need to be resumed.
Tuesday. The Senate confirmed Warsh by 54 votes to 45. For context: Jerome Powell was confirmed 84 to 13 in 2018 and 80 to 19 in 2022. The 54-to-45 vote is not a signal about Warsh's competence. It is a signal about the political temperature of monetary policy. Few Fed Chairs have taken office with less bipartisan goodwill.
Wednesday. The April CPI print landed at 3.8 percent, against a Bloomberg consensus of 3.4 percent. Energy accounted for 1.1 percentage points of the surprise. Shelter services added another 0.4 percentage points. The headline number was not a statistical accident — it was a mechanical output of WTI closing above 95 dollars for six consecutive weeks. The bond market responded immediately: the 10-year yield moved from 4.38 percent to 4.52 percent in the session.
Thursday. Cerebras Systems priced at 185 dollars per share on Wednesday evening and opened at 350 on Thursday morning. By the close it had settled at 311 dollars — a 68 percent gain on day one, making it the largest IPO of 2026 and raising 5.55 billion dollars. The company makes chips. Not chips in the conventional sense: its Wafer Scale Engine is a single piece of silicon the size of a dinner plate, 58 times larger than Nvidia's B200 and packed with 4 trillion transistors. The OpenAI compute deal, worth over 10 billion dollars in confirmed commitments, provides near-term demand visibility that most semiconductor startups would consider a fantasy. The market is paying approximately 100 times trailing sales for the thesis that wafer-scale architecture eventually makes the GPU cluster obsolete. That is an audacious bet. The day-one performance suggests the market is not yet asking what happens if the thesis is wrong. Also on Thursday: the Geneva talks between US and Chinese trade negotiators concluded without a framework agreement. Both sides described the sessions as "constructive" — the diplomatic equivalent of saying the meal was fine when you are not ordering dessert. The talks will continue. The tariffs will continue.
Friday. WTI Crude settled at 105.42 dollars, up 6.8 percent on the week, as shipping insurance costs across Hormuz-adjacent routes continued to rise sharply. The psychological significance of 100 dollars a barrel is not the number itself — it is what 100-dollar oil does to PCE, the Fed's preferred inflation measure, over the following six to eight weeks. Warsh will be reading those numbers on his first day in the building.
The dominant action for a multi-asset portfolio this week is not directional — it is about your sensitivity to the Warsh reaction function. If you are long growth at 20x-plus earnings and you have no hedge on a 25-basis-point hike scenario, the CPI print is telling you something. The energy position is working. The question is whether the energy-inflation loop is starting to destroy the multiple on everything else. Review your rate sensitivity before the next FOMC meeting. The Warsh Holds scenario at 45 percent probability does not mean the risk is remote.
The question this week is not whether AI capital expenditure is real. Three full earnings cycles have settled that. The question — the one the Citrini-Citadel score now tracks at 41 to 35 — is whether anyone other than the builders is profiting from it. And this week gave that question its sharpest frame yet.
A new Federal Reserve Chair arrived into a 3.8 percent CPI reading driven partly by energy prices that are themselves partly driven by the power demand of the data centres that are themselves driven by the AI infrastructure build. The loop is now closed. The capital expenditure that was supposed to be deflationary in the long run is inflationary in the medium run because it consumes electricity, water, and specialised materials at a rate that the Hormuz disruption has made measurably more expensive. Warsh's Taylor Rule problem and the AI capex cycle are the same problem, priced in different markets.
Three quarters ago, when this publication first flagged the enterprise return-on-investment gap, 95 percent of enterprise AI deployments were reporting zero measurable return. That figure has not materially improved. Citadel's position — that the infrastructure is running ahead of an application layer that has not yet materialised — has the empirical data behind it. Citrini's counter remains structural: the companies building the infrastructure win regardless of whether the applications generate returns for their customers. The hyperscalers are simultaneously the picks-and-shovels business and the mining company. Citrini is winning the score. But the score updates on 26 May, when Nvidia reports first-quarter results. If the forward guidance implies continued data centre build-out at current pace, the infrastructure thesis holds. If guidance signals any deceleration, Citadel's ROI-gap argument finds its first hard footing.
Thursday's Cerebras Systems IPO is the most direct expression of this infrastructure thesis the market has offered all year. At 185 dollars, priced the night before opening, it raised 5.55 billion dollars. By Thursday's close it had added 68 percent. The Wafer Scale Engine — a chip 58 times the physical size of Nvidia's B200 — is a direct architectural bet that the GPU cluster model eventually hits a wall. The OpenAI compute deal — confirmed at over 10 billion dollars, with a broader arrangement under discussion — is not a product endorsement. It is a hedge by the world's most consequential AI lab against the possibility that its compute costs remain captive to a single supplier. Wall Street is paying 100 times trailing sales for the outcome where that hedge proves to have been early rather than wrong. The valuation is audacious. The strategic logic is not.
CuspAI's Series A this week adds a structural layer to this picture. The industrial strategic investors backing it are not placing a bet on AI adoption timing. They are making a bet that the materials constraints feeding the AI infrastructure stack — gallium, germanium, specialty substrates, ultra-pure quartz — will remain binding for long enough to justify funding the company that might eventually solve them. When major industrial buyers make that move, they are telling you something about the duration of the build that no quarterly earnings call can.
This will take four minutes. Paste the following into Claude: "A new Fed chair has just inherited a 3.8 percent CPI partly driven by energy costs that are partly driven by AI data centre power demand. Walk me through the causal chain from hyperscaler capital expenditure to the Taylor Rule gap — quantifying each step where possible. What is the strongest counter-argument that this chain is coincidental rather than structural?"
What Anthony found: The model traced the chain with reasonable precision — data centre power demand now accounts for approximately 2 to 3 percent of US grid consumption and is projected to reach 8 percent by 2030, which at current natural gas prices adds roughly 0.2 to 0.3 percentage points to the energy services component of CPI. The counter-argument it produced was more interesting: the chain is real but the causation runs the other way. Cheap energy enabled the AI build. Expensive energy will accelerate the transition to more efficient chips. The inflationary pressure is self-limiting. Whether that self-limiting mechanism arrives before Warsh has to make a rate decision is the actual question.
One observation: The market is pricing the AI infrastructure story and the inflation story as separate trades. The analytical edge is in understanding that they are not.
Hormuz as a price negotiation. WTI at 105.42 dollars is not a supply emergency. It is an opening bid in a negotiation that has not yet formally started. The disruption to Hormuz transit has been real — vessel counts through the strait were down 23 percent in the rolling four-week average as of Thursday — but the market is pricing a structural premium, not an acute shortage. The difference matters: acute shortages resolve quickly and the premium collapses; structural premiums persist until the underlying negotiation concludes.
Shipping insurance costs for Hormuz-adjacent routes rose sharply again this week, reflecting what insurers are treating as an elevated and sustained risk environment rather than a transitory disruption. When insurance pricing moves at this pace, the cost of credit for energy logistics firms moves next, and the cost of physical supply chains — including the materials supply chains feeding semiconductor manufacturing — follows six to eight weeks later.
The complacency pattern. Markets have been wrong about major supply shocks at exactly these moments before. In July 1914 they priced a negotiated resolution. In October 2007 they priced a contained housing correction. In February 2020 they priced a localised health event. In each case the mechanism was the same: the disruption was visible and the downstream consequences were not priced until they arrived in the data. The Hormuz disruption has been visible since February. The inflation consequences are beginning to show. What is not yet priced is the persistence — this is not a transitory event but a structural one, with a negotiating timeline that has no diplomatic catalyst currently visible. This is the analytical gap between where markets are today and where they may be forced to move if the data continues to arrive as it has been arriving.
UK Gilts versus US Treasuries. The 16 basis point spread between UK gilts at 4.68 percent and US 10-year Treasuries at 4.52 percent reflects three separate forces: UK services inflation running above the Bank of England's tolerance at 5.1 percent, a UK fiscal trajectory that has not found its equilibrium, and sterling weakness that is making imported inflation stickier than the Bank's models predicted. Developed market bond divergences of this scale — when both economies are supposedly in the same cyclical phase — are a warning that the phase alignment is illusory.
Geneva and the pace of separation. The US-China trade talks produced language but no framework. Supply chains have now had fourteen months to adapt to the post-2025 tariff landscape and they are doing so in ways that make a reversal progressively more expensive. The capital expenditure decisions that went into Vietnam, India, and Mexico last year are not reversible in twelve months. The political negotiation is moving at a diplomatic pace; the economic separation is moving at a capital allocation pace. They are not converging.
EM currency divergence. The Turkish lira fell 30.9 percent against the dollar this week — a continuation of the structural deterioration in Turkey's external position. The South African rand strengthened 8.7 percent, supported by commodity export receipts from platinum and palladium at elevated prices. These two moves are not related by underlying cause: Turkey's lira collapse is a domestic policy story; the rand's strength is a commodity story. The error is treating them as a single "EM" signal. They are not.
Why the negotiations are stalled. The US position requires Iran to simultaneously surrender uranium enrichment and reopen the Strait in exchange for sanctions relief and access to frozen assets. Iran's strategy is to resist this sequencing — to accept a phased deal where the Strait reopens first and uranium is negotiated separately, securing the economic benefit before making the harder concession. Iran needs only its oil revenues to continue flowing at a survivable rate, which they are barely managing. The consequence is a diplomatic timeline considerably longer than equity markets are pricing. A partial framework is unlikely before early July at the earliest. Full normalisation, if it comes, is a midsummer story at the earliest. Economic pain from the disruption will persist well beyond any framework announcement.
The Taiwan dimension. The secondary geopolitical risk is not where most portfolios are hedged. The relevant scenario is not military invasion of Taiwan — current assessments suggest that is not a near-term plan. It is something subtler: customs assertion, with Coast Guard vessels declaring authority over Taiwanese trade routes, forcing the United States to escalate against a non-military action or accept a creeping fait accompli. This scenario matters because US precision weapons stockpiles are depleted following the Iran campaign, creating a temporary window where such an assertion carries reduced response risk. The ultimate strategic prize is clear: any restriction on TSMC's semiconductor exports changes the global AI infrastructure map immediately. Equity markets are not pricing this tail. They should at least be aware of its shape.
Three scenarios for Hormuz:
This week's data environment is dense with things getting worse. CPI at 3.8 percent. WTI above 100 dollars. A Fed chair confirmed by a margin that suggests the political system is more fractured than the equity market has priced. None of these are trivial concerns.
Here is what the tape is obscuring. In 1900, global average life expectancy at birth was approximately 32 years. In 2024, it is 73 years. That is not a rounding error. That is a doubling and a bit more, achieved across a period that included two world wars, the 1918 influenza pandemic, the Great Depression, the Cold War, multiple regional conflicts, the HIV epidemic, and every inflationary shock in between — including several that were considerably worse than 3.8 percent CPI.
The compounding effect of that improvement — in nutrition, in sanitation, in medicine, in the fundamental organisation of economic activity — does not show up in a weekly tape. It is not in the CPI print. It is not in the Warsh confirmation vote tally. It is in the 40 additional years of life that an average human being today experiences compared to an average human being in 1900. That is the context within which this week's concerns should be held.
This is not a reason to dismiss the inflation data or the Hormuz risk or the Taylor Rule arithmetic. It is a reason to hold those concerns in proportion. The compounding of human progress is slower than a Fed decision and faster than a geopolitical cycle. It is not visible week to week. It is visible decade to decade. And over any decade you care to measure in the past two centuries, the direction has been the same.
Serra Verde is a Brazilian rare earth producer that completed its NYSE listing this week via a Special Purpose Acquisition Company merger. It is the only rare earth company in the Western hemisphere with a producing asset — the Verde Grande deposit in Bahia state — and an in-house separation facility. The thesis connects directly to the week's dominant macro theme: if the Hormuz disruption is the energy price story, the US-China separation at Geneva is the materials price story. China controls approximately 85 percent of global rare earth processing. Serra Verde is one of three assets in the world that could credibly serve as an alternative source for the heavy rare earths — dysprosium, terbium, holmium — that are critical for permanent magnets in EV motors and wind turbines.
The historical parallel is the cobalt price cycle of 2016 to 2018, when the DRC's 63 percent share of global cobalt production became visible to the EV supply chain simultaneously. Cobalt prices tripled in 18 months before alternative supply and battery chemistry changes brought them back down. The rare earth cycle will have a similar dynamic: a long lead-time discovery phase, followed by rapid price dislocation when the demand inflection is undeniable, followed by supply response. Serra Verde is attempting to position before the dislocation. The Special Purpose Acquisition Company route to listing raises governance questions; the Bahia deposit environmental licensing history raises compliance questions. Both are manageable at the current price if the strategic thesis is correct.
Thesis breaker: CADE, the Brazilian competition authority, challenges the Serra Verde acquisition structure; or the STF, Brazil's supreme court, revisits land use permissions in the Bahia extraction zone. Both are tail risks, not base cases.
Anthony Rosenthal analysis score: 6.20 / 10 — WATCHLIST / AMBER. Strategic case is compelling; execution risk, governance discount from Special Purpose Acquisition Company structure, and Brazilian political risk combine to make this a watch rather than a commit at the opening price. Four-week score date: Week 22.
Every company that appears in On the Radar has been run through a structured eight-section analytical framework before publication. The score shown is a weighted composite across ten dimensions, each scored out of ten. Nothing appears here without this analysis being completed first.
Even-week edition: all 12 strategies refreshed with current yields, spread moves, and any thesis changes following the CPI 3.8 percent print and the re-inversion of the yield curve.
Week in review. The 3.8 percent CPI print has moved yields uniformly higher across the ERDR universe — all 12 strategies are now yielding 15 to 20 basis points more than the Week 17 dashboard. The re-inversion of the yield curve to 4.52 percent on the 10-year versus 4.62 percent on the 2-year changes the case for duration-heavy strategies and improves the case for floating-rate and short-duration structures. Active income funds with Lombard overlay and senior secured leveraged loans are the primary beneficiaries of the week's moves.
| # | Strategy | Yield | Spread vs IG | WoW | Thesis | Action |
|---|---|---|---|---|---|---|
| 1 | Active income fund + Lombard | 8.2% | +320 to 340bps | +18bps | Floating rate component benefits from re-inversion; Lombard overlay adds leverage efficiency. CPI print confirms the carry case. | Add |
| 2 | IG/split-rated CLO tranches | 6.8% | +180 to 200bps | +15bps | Structural protection from CLO waterfall holds through the CPI shock. Post-CPI entry point constructive for a 6-month hold. | Add |
| 3 | Listed infrastructure debt/equity | 5.9% | +90 to 110bps | +12bps | Duration sensitivity creates short-term headwind as 10Y re-inverts. Long-run inflation linkage of infrastructure revenues is supportive. Hold through the rate volatility. | Hold |
| 4 | Business Development Companies | 9.1% | +410 to 430bps | +20bps | Floating rate loans repriced higher with base rate. Credit quality holding in the upper mid-market. Watch for rising default rates in leveraged buyouts from the 2021 to 2022 vintage. | Watch |
| 5 | Agency mortgage REITs | 10.4% | +480 to 500bps | +22bps | Yield attractive but duration risk elevated as 10Y moves. Prepayment speeds slowing — lock-in effect benefits carry but creates convexity risk if rates reverse sharply. | Watch |
| 6 | Senior secured leveraged loans | 8.7% | +370 to 390bps | +19bps | Floating SOFR-plus structure reprices immediately as base rate holds or rises. Senior secured position provides downside protection. The week's best risk-adjusted response to the CPI print. | Add |
| 7 | Preferred shares and hybrid capital | 7.3% | +250 to 270bps | +17bps | Rate sensitivity of fixed-rate preferreds is the issue. Hybrid capital with reset mechanisms less affected. Watch the Warsh reaction function — a hike scenario reprices this category materially lower. | Watch |
| 8 | Real asset royalties | 7.8% | +300 to 320bps | +16bps | Commodity royalties benefit directly from WTI 105.42 dollars and elevated base metals. The energy royalty component is the strongest performer in the universe this week. | Add |
| 9 | EM hard-currency sovereign carry | 8.9% | +390 to 410bps | +18bps | Dollar strength from CPI print pressures EM positioning. Commodity-exporting EM sovereigns better positioned than importers. Watch selectively: Brazil and South Africa outperform Turkey and South Asia. | Watch |
| 10 | High-yield municipal bonds | 5.4% | +60 to 80bps | +10bps | Tax-exempt yield less affected by the CPI shock in after-tax terms for top-bracket holders. Duration risk present but less acute than taxable equivalents at this yield level. | Hold |
| 11 | Private credit direct lending | 10.8% | +560 to 580bps | +20bps | Illiquidity premium intact. Floating rate structure benefits from the base rate environment. Covenant documentation from 2022 to 2024 vintages generally tighter than the 2019 to 2021 cohort. | Hold |
| 12 | Trade and supply chain finance | 7.1% | +230 to 250bps | +14bps | Short-duration 30 to 90 day structures insulated from rate volatility. Geneva stalemate adds friction to China-US trade flows but creates opportunity in alternative route financing. | Hold |
Action ratings: Add / Hold / Watch / Reduce. All yields are indicative, not guaranteed. Appendix D contains the full methodology note.
The thing about the Taylor Rule is that it was designed to be a guide, not a governor. John Taylor proposed it in 1993 as a way of describing what well-functioning central banks had been doing, not as a mandate for what they should do. It is, at its core, a backwards-looking description dressed in the clothes of a forward-looking prescription.
Which means that Kevin Warsh inherits not just the Taylor Rule gap but the question of what the Taylor Rule means in a world where energy prices are set partly in the Strait of Hormuz rather than in Cushing, Oklahoma. The original formulation did not include a variable for geopolitical premium. It was built for a world where supply was, broadly, a function of price incentives. The world of 2026, in which 23 percent of global oil transit has been disrupted by a combination of military posturing and insurance market mechanics, is a world the Taylor Rule was not designed to navigate.
Warsh knows this. The question is whether his first press conference will say so, or whether it will describe a 3.8 percent CPI as a domestic monetary failure rather than a partially exogenous energy event. The language matters more than the rate decision. And language, as any central banker who has read their Greenspan transcripts will tell you, is the actual instrument.
Meanwhile, somewhere in Cambridge, Chad Edwards is running a language model on experimental battery data and finding things that took previous generations of chemists a decade to find. The model does not know about the Taylor Rule. It does not know about the Hormuz premium or the 54-to-45 vote. It knows about molecular bond energies and crystalline lattice configurations and the failure modes of previous electrode materials. It is doing its job with complete indifference to the macroeconomic environment. This, on reflection, seems correct.
A footnote on scarcity, since this week has had a lot to say about it. Cerebras listed at 185 dollars and closed at 311. The AI market is paying 100 times sales for the thesis that wafer-scale silicon displaces GPU clusters at the frontier. Across town — metaphorically — the average NBA franchise carries debt of under 10 percent of its asset value, compounding at rates that have consistently exceeded the technology sector over the past decade. The scoreboard keeps score in the long run. What is truly scarce does not stay cheap forever.
25 assets · 1 Jan 2026 baselines · Week 18 close
| Asset | Baseline (1 Jan) | Current | YTD % |
|---|---|---|---|
| Baltic Dry Index | 1,877 | 3,151 | +67.87% |
| WTI Crude Oil | $63.20 | $105.42 | +66.81% |
| USD/TRY | 35.38 | 45.49 | +28.50% |
| Nikkei 225 | 51,832 | 61,409 | +18.48% |
| Nasdaq 100 | 25,200.50 | 29,125 | +15.57% |
| Copper ($/lb) | $5.68 | $6.50 | +14.40% |
| Russell 2000 | 2,481.91 | 2,793 | +12.55% |
| S&P 500 | 6,845.50 | 7,408.50 | +8.22% |
| Silver ($/oz) | $70.62 | $75.75 | +7.29% |
| Gold ($/oz) | $4,341 | $4,547.89 | +4.76% |
| FTSE 100 | 9,945 | 10,373 | +4.27% |
| HYG (High Yield) | $78.18 | $79.49 | +1.71% |
| Euro Stoxx 50 | 5,739 | 5,823.50 | +1.45% |
| MSCI EM | 1,595 | 1,600 | +0.30% |
| Hang Seng | 26,370 | 26,389 | +0.19% |
| DAX | 24,571 | 24,456 | -0.34% |
| Swiss SMI | 13,248 | 13,177 | -0.54% |
| LQD (IG Corp) | $109.22 | $107.95 | -0.98% |
| Nifty 50 | 24,427 | 23,690 | -2.99% |
| AGG (US Agg Bond) | $102.10 | $98.89 | -3.19% |
| USD/ZAR | 17.55 | 16.68 | -4.96% |
| Bitcoin (BTC) | $87,850 | $79,611 | -9.38% |
| TLT (Long Treasury) | $94.27 | $83.80 | -11.11% |
| Natural Gas | $3.53 | $2.96 | -15.77% |
| Ethereum (ETH) | $2,967 | $2,261 | -23.79% |
Every company that has appeared in On the Radar remains tracked here until its formal four-week score date. A company moves from On the Radar to this appendix when there is no new catalyst that week — the analytical call is intact, but there is nothing fresh to add.
| Company / Ticker | Entry / Week | Current | Return | Original Thesis | Score Date |
|---|---|---|---|---|---|
| Talen Energy NYSE: TLN |
$361.01 Week 15 |
~$401.00 | +11.1% | Nuclear power co-location opportunity for data centre operators; unique PJM Interconnection grid position; hyperscaler offtake discussions ongoing. | Week 19 |
| Venture Global LNG NYSE: VG |
$13.08 Week 16 |
~$12.20 | -6.7% | Highest uncontracted volume ratio among US LNG exporters; Ras Laffan capacity offline 3 to 5 years; spot price premium versus long-term contracts. | Week 20 |
| MP Materials NYSE: MP |
$67.21 Week 17 |
~$71.50 | +6.4% | Only US-based rare earth producer with Mountain Pass deposit; China controls 85 percent of global rare earth separation; strategic decoupling trade. | Week 21 |
| Serra Verde Materials NYSE: USAR |
~$21.00 Week 18 |
$21.00 | — | Only Western hemisphere rare earth producer with in-house separation; strategic decoupling from Chinese rare earth processing; Serra Verde deposit in Bahia, Brazil. | Week 22 |
| Index | Level | WoW | YTD |
|---|---|---|---|
| S&P 500 | 7,408.50 | +1.2% | +8.2% |
| Nasdaq 100 | 29,125 | +0.8% | +15.6% |
| Russell 2000 | 2,793 | +0.4% | +12.6% |
| Dow Jones Industrial Average | 42,350 | +0.4% | +6.2% |
| S&P 500 Energy | 891 | +4.1% | +23.6% |
| S&P 500 Technology | 3,420 | -0.5% | +8.3% |
| Instrument | Yield | WoW Change |
|---|---|---|
| US 2Y Treasury | 4.62% | +24bps |
| US 5Y Treasury | 4.55% | +17bps |
| US 10Y Treasury | 4.52% | +14bps |
| US 30Y Treasury | 4.97% | +12bps |
| UK Gilt 10Y | 4.68% | +11bps |
| German Bund 10Y | 2.71% | +8bps |
| Bloomberg US Agg | -1.8% YTD | -0.6% WoW |
| Commodity | Price | WoW | YTD |
|---|---|---|---|
| WTI Crude Oil | $105.42/bbl | +6.8% | +66.8% |
| Brent Crude | $107.20/bbl | +6.4% | +39.8% |
| Natural Gas (Henry Hub) | $2.96/MMBtu | -2.1% | -15.8% |
| Gold | $4,547.89/oz | +1.4% | +4.8% |
| Silver | $75.75/oz | +0.8% | +7.3% |
| Copper | $6.50/lb | -1.1% | +14.4% |
| Wheat (CBOT) | $6.14/bu | -0.4% | +8.2% |
| Pair | Rate | WoW | YTD |
|---|---|---|---|
| EUR/USD | 1.0882 | +0.6% | +5.1% |
| GBP/USD | 1.2730 | +0.3% | +2.8% |
| USD/JPY | 154.40 | +0.9% USD stronger | +3.2% USD stronger |
| USD/CHF | 0.8960 | +0.2% USD stronger | +1.1% USD stronger |
| DXY | 103.8 | -0.4% | -3.3% |
| Pair | Rate | WoW | YTD |
|---|---|---|---|
| USD/TRY (Turkish Lira) | 38.40 | +30.9% USD stronger | -41.2% TRY |
| USD/ZAR (South African Rand) | 17.85 | +8.7% ZAR stronger | +4.2% ZAR |
| USD/BRL (Brazilian Real) | 5.62 | +1.4% | -3.8% BRL |
| USD/INR (Indian Rupee) | 84.20 | +0.3% | -1.1% INR |
| USD/CNY (Chinese Renminbi) | 7.24 | Flat | -0.6% CNY |
| Indicator | Level | WoW | Context |
|---|---|---|---|
| VIX (Equity Vol) | 18.2 | +1.7pts | Elevated; CPI shock absorbed but not digested |
| MOVE Index (Rate Vol) | 112 | +8pts | Highest since Warsh confirmation week; rate uncertainty elevated |
| HY-IG Credit Spread | +310bps | +12bps | Widening modestly; no credit event, but Warsh uncertainty repricing risk |
| IG Credit Spread | +110bps | +5bps | Contained; investment grade markets functioning normally |
| SKEW Index | 138 | +4pts | Tail risk buying elevated; downside protection being purchased |
| Indicator | Latest | Prior | Consensus |
|---|---|---|---|
| US CPI YoY | 3.8% | 3.5% | 3.4% |
| US Core CPI YoY | 3.4% | 3.2% | 3.1% |
| US PCE YoY (prev) | 3.1% | 2.9% | — |
| US Unemployment Rate | 4.1% | 4.0% | 4.0% |
| US ISM Manufacturing | 48.7 | 49.2 | 49.0 |
| Michigan Consumer Sentiment | 61.4 | 65.2 | 63.0 |
| Hormuz Vessel Count (4W avg) | -23% | -19% | — |
| Asset | Price | WoW | YTD |
|---|---|---|---|
| Bitcoin (BTC) | $79,611 | -1.4% | -9.4% |
| Ethereum (ETH) | $2,261 | -1.8% | -23.8% |
| Solana (SOL) | $182 | +3.2% | -8.4% |
| Total Crypto Market Cap | $3.28 trillion | +1.4% | -4.2% |
| BTC Dominance | 58.3% | +0.4pts | +5.1pts YTD |
| Date | Event | Significance |
|---|---|---|
| 20 May | First Warsh Fed Statement (FOMC Minutes) | Critical — language on reaction function sets the tone |
| 22 May | US PCE Deflator (April) | High — the Fed's preferred inflation measure; will it confirm CPI? |
| 23 May | UK CPI (April) | Moderate — UK inflation trajectory relevant to gilt divergence |
| 26 May | Nvidia Earnings (Q1 FY2027) | High — AI capital expenditure validation; hyperscaler demand signal |
| 28 May | US GDP Revision (Q1 2026) | Moderate — confirm or revise the growth picture |
| 30 May | Week 22 — 2026 Thesis Check-In | Internal — scheduled check-in on Year of the Repricing thesis |