“A ceasefire is not peace. It is simply the market’s permission to forget what frightened it.”
— Anthony Rosenthal, Weekly Market Pulse, April 2026Macro
Week 13 — Record. Conditional.
The Strait of Hormuz reopened to commercial traffic on Friday. The S&P 500 closed at a record 7,126. Oil fell more than 11%. The US naval blockade remains in place. These facts are not contradictory. They are the same story told from different time horizons.
| Asset Class | Signal | Change | Note |
|---|---|---|---|
| Equities | ▲ Improved | Record 7,126 on conditional catalyst. Built on a single session, triggered by a single announcement. Watch Monday open. | |
| Fixed Income | ■ Stable | 10Y at 4.26%, -5bps on the week. Curve normalising. Kevin Warsh hearing Tuesday is the variable that reprices everything. | |
| Commodities | ▲ Improved | WTI -11.45% WoW but +23.3% YTD. Copper +50% YTD. Energy easing; industrial metals structurally bid. | |
| Digital Assets | ■ Stable | BTC $75,864, -13.5% YTD. Tactical bounce on risk-on. Not a structural conviction buy yet. |
Last week we flagged the ceasefire as conditional and warned that the naval blockade remaining “in full force” meant the Strait reopening was a headline, not a resolution. That proved correct — WTI has fallen but Brent remains 32.9% above its January baseline and shipping insurers have not removed their force majeure clauses. We did not anticipate the magnitude of the single-session rally (+4.54% S&P) or the Bloom Energy/Oracle announcement as a simultaneous catalyst. Both accelerated the risk-on move faster than the underlying data warranted.
Whether Kevin Warsh’s Senate testimony on Tuesday signals rate flexibility or structural hawkishness. That single hour of testimony will reprice the forward curve, the equity multiple, and gold simultaneously. Everything else this week is downstream of it.
Analysis
Three things must remain true for this rally to hold.
Every point on the S&P 500’s chart above 7,000 rests on three assumptions. If any one of them fails, the others do not simply adjust — they tend to collapse together.
Assumption 1: The Strait stays open. Iran’s Foreign Minister declared the Strait “completely open” on Friday. The US President, in the same news cycle, confirmed the naval blockade “remains in full force.” These two statements are not inconsistent. The Strait is commercially open while the blockade remains in place. The difference matters enormously to anyone pricing energy beyond the front month. The rally is pricing the opening. The physical market has not yet cleared the blockade.
Assumption 2: Kevin Warsh signals continuity. The Fed Chair-in-waiting testifies before the Senate Banking Committee on Tuesday. Warsh, whose nomination in January triggered the sharpest single-day decline in gold and silver prices in decades, is viewed as more hawkish than current Fed leadership. If he signals a preference for a structurally higher rate floor, the equity multiple faces compression. Markets are currently pricing a 50-50 chance of a 25-basis-point cut by December. That probability collapses or expands on Tuesday morning.
Assumption 3: AI earnings remain structural. TSMC reported quarterly profit +58%. Bloom Energy surged 23.98% on a 2.8-gigawatt Oracle deal. These are earnings moves, not sentiment moves. But every rational investor must ask: at what point does the AI capital expenditure build exceed the near-term addressable return? The answer is not yet. But the distance between “not yet” and “now” is measured in quarters, not years.
Base case — Flexibility signalled: Warsh acknowledges the energy shock as a deflationary drag and declines to pre-commit to a higher rate floor. Yield curve targets: 2Y holds 4.05–4.25%, 5Y holds 3.95–4.15%, 10Y holds 4.15–4.35%. Equity multiples remain intact. A December 25bp cut stays in play. Probability: 55%.
Stress case — Structural hawkishness: Warsh signals a preference for a rate floor above 4.5% regardless of energy dynamics. Yield curve targets: 2Y reprices toward 4.65–4.85% (+30–50bps), 5Y toward 4.45–4.60%, 10Y toward 4.60–4.75% within the session. The equity multiple embedded in 7,126 faces compression of 150–200 index points. A bear-flattening move of this magnitude would reprice the entire forward strip; gold and silver pull back sharply — as they did at his nomination in January. Probability: 45%.
| Asset Class | View | Rationale |
|---|---|---|
| Equities | Amber — Conditional positive | Record but fragile. Built on geopolitical relief, not structural earnings acceleration. |
| Fixed Income | Amber — Stable | Curve normalising. Warsh hearing is the decisive event for rate path pricing. |
| Commodities | Green — Structural | Copper +50%, silver +12.6% YTD on genuine AI demand and supply constraints. Energy conditional. |
| Digital Assets | Amber — Tactical | Risk-on bounce. BTC -13.5% YTD; ETH -20.8% YTD. Notably, institutional accumulation in ETH is running at approximately double the rate of BTC. Price down, accumulation up: this is either a classic liquidity trap (forced sellers overwhelming patient buyers) or the setup for a sharp mean-reversion when the overhang clears. The divergence is structural, not explained by sentiment. |
| Overall Risk | Amber — Improving | Hope has improved. The data has not yet confirmed it. |
Entrepreneur of the Week
The machine that reads the mind — and what it says about the next frontier of human potential.
In January 2024, a 29-year-old man named Noland Arbaugh, paralysed from the shoulders down following a diving accident, sat in front of a computer screen and moved the cursor with his thoughts. He was the first human to receive a Neuralink implant. Within days, he was playing chess online. Within weeks, he was gaming for eight hours at a stretch. He described the experience simply: “It is like using the Force.”
Neuralink was founded in 2016 by Elon Musk and a group of neuroscientists with a thesis that most people found implausible at the time: that a brain-computer interface small enough to be implanted through a keyhole procedure, safe enough for everyday use, and sophisticated enough to decode neural signals in real time was not decades away but years. The scientific community was sceptical. History has a way of being unkind to the sceptics of useful things.
The device itself — branded the N1 implant — is a chip roughly the size of a large coin, inserted by a precision-engineered surgical robot into the motor cortex. It records electrical signals from neurons and translates them into digital commands. The bandwidth is extraordinary: the current version reads from over a thousand electrodes simultaneously, a leap over anything that existed in clinical use when Neuralink began its work. The second device, called Blindsight, is designed to restore rudimentary vision to those who have lost it entirely — even those born blind. The ambition is not incremental.
Musk’s thesis has always been broader than medical device. He has argued, consistently and with characteristic directness, that artificial intelligence will surpass human cognitive ability within this decade, and that the only rational response is to augment human intelligence rather than compete with it. Neuralink is not, in his framing, a medical company. It is a hedge. The brain-computer interface is the merger that preserves the balance of power between human and machine.
What makes Neuralink relevant to readers of this publication is not the science alone. It is the business model that the science enables. The addressable market for neural interfaces begins with medical necessity — paralysis, blindness, severe depression, Parkinson’s — and expands, as the technology matures and regulatory frameworks follow, toward cognitive enhancement for healthy individuals. That is a market of potentially billions rather than millions. The question every investor should be asking is not whether brain-computer interfaces will exist at scale. They will. The question is who will own the standard.
Right now, Neuralink owns the most compelling clinical data, the most advanced surgical robot, and the most recognisable brand in a field where brand recognition is worth more than it usually is. It also has a founder whose attention is, to put it diplomatically, distributed across several of the most consequential technology bets of the century. That is both the greatest asset and the greatest risk in the same sentence.
But here is the counter-intuitive dimension that the medical framing obscures. Noland Arbaugh — paralysed, implanted, gaming for eight hours at a stretch — was not performing at a disabled person’s level. He was out-performing healthy users on the same platforms. The N1 implant routes around the bottleneck of the human hand. In the near term, this is a story about restoring function. In the medium term, it is a story about something more uncomfortable: if an augmented human can consistently outperform an unaugmented one, then “unassisted” becomes a competitive disadvantage rather than the baseline. That is not a distant prospect. It is already the implicit logic of every productivity software company in Silicon Valley, made biological. Neuralink is simply the version that does not require you to look at a screen.
The entrepreneur’s job is to see around corners. Musk saw this one clearly: the human brain, unassisted, is the bottleneck. Neuralink is the cable. The question worth sitting with is what happens to everyone who cannot afford the connection.
The ESG implications of augmented human labour have not yet entered mainstream institutional analysis. They should. If an N1-augmented worker can demonstrably out-perform an unaugmented peer — in data entry, surgical precision, or complex decision-making — the competitive logic of the labour market begins to select for augmentation. The augmented individual gains. The unaugmented majority, for whom the procedure is either unavailable, unaffordable, or culturally inadmissible, faces structural disadvantage. This is not a sci-fi scenario. The logic is already visible in lower-stakes technologies: algorithmic trading versus manual, AI-assisted diagnosis versus unaided. Neuralink is the extreme case of a dynamic already in motion.
The social stratification of cognitive labour is an underwritten risk in the standard ESG framework. The S-pillar of ESG — social — has historically focused on labour practices, supply chain conditions, and community impact. It has not yet developed a vocabulary for a world in which cognition itself is a purchasable upgrade. Investors with a five-year horizon should build one now, before the market forces them to.
Weekly Review
Monday relief, Tuesday conviction, Friday record.
Monday, 13 April. The ceasefire from Week 12 held through the weekend. Markets opened cautiously. Oil futures settled. The question hanging over every trading desk was whether last week’s surge was the high or merely the first instalment.
Tuesday, 14 April. Bloom Energy surged 23.98% after confirming an expanded partnership with Oracle: 2.8 gigawatts of solid oxide fuel cell systems to power next-generation AI data centres. The deal was more than a corporate announcement. It was a reframing — the energy story shifted, in a single session, from crisis to infrastructure. Investors who had sold energy stocks on the ceasefire news spent Tuesday buying them back.
Wednesday, 15 April. Natural gas hit a 17-month low at $2.67 per MMBtu, a reminder that not every energy commodity responds to geopolitical narrative. Copper pressed through $6.10 per pound, extending its year-to-date gain to 50% from our January baseline. Institutional attention began turning to silver, where China’s export limits on refined metal have created a supply deficit that is starting to appear in the forward curve.
Thursday, 16 April. TSMC reported its first quarter results: profit up 58% year on year. Five consecutive quarters of double-digit earnings growth for the global AI hardware complex. The semiconductor index closed at a fresh all-time high. For those who had been waiting for a sign that the AI capital expenditure cycle was something other than narrative, this was it.
Friday, 17 April. Iran’s Foreign Minister Abbas Araghchi declared the Strait of Hormuz “completely open to all commercial traffic.” WTI fell 11.45%. The S&P 500 surged. By 4pm Eastern, the index had crossed 7,100 for the first time in history. Elsewhere: Netflix fell 9.72% after its co-founder announced retirement and second-quarter guidance disappointed. Live Nation fell 6.3% after a jury found the company had operated an illegal concert monopoly. Both stories were true, important, and almost entirely ignored. Everyone was watching oil.
Risk Assessment
Three structural trades and one growing concern.
Froth Score: 6.2 / 10 ▲ from 5.8 last week
The AI Power Premium. Bloom Energy’s gain of nearly 24% in a single week on a single customer announcement is a reminder that markets can price structural change faster than structural change can arrive. The underlying thesis is sound: AI data centres need reliable, behind-the-meter power that the grid cannot consistently supply. The question is whether the current valuation of the AI power complex already anticipates several years of deployment. The signals suggest at least two years are in the price.
Copper at +50% YTD. No commodity produces a 50% year-to-date gain in a non-crisis environment without attracting scrutiny. Copper’s rise combines genuine AI infrastructure demand, tariff-driven front-running of US import substitution, and significant speculative positioning from trend-following funds. The structural story is real. The positioning risk is high. These are not mutually exclusive.
The “Complete” Strait. Markets traded Friday as if the Strait of Hormuz had been definitively resolved. It has not. “Completely open” and “blockade in full force” are simultaneously true, which means one of them will need to yield. Energy markets have priced the first statement. If the second asserts itself, WTI returns toward $95 within days.
Silver’s Structural Deficit. China’s export limits on refined silver, effective 1 January 2026, have removed approximately 70% of global discretionary supply from the open market. At $80.75 per ounce, silver is up 12.6% year to date. The institutional target of $100 remains below where copper’s YTD trajectory would imply it should trade on a relative-demand basis. This is either a lag in recognition or a signal that the supply story is less acute than reported. Both possibilities deserve attention.
Macro Intelligence
Four developments that moved faster than the week’s consensus.
Bloom Energy / Oracle: 2.8 gigawatts of solid oxide fuel cell deployment confirmed across Oracle’s AI data centre network. Behind-the-meter power has moved from pilot project to standard specification for hyperscale computing. The implications for grid utilities and independent power producers are significant and are only beginning to be priced.
Kevin Warsh Hearing, Tuesday 10am EST: The Fed Chair-in-waiting testifies before the Senate Banking Committee. No Federal Reserve transition in modern history has occurred during an active geopolitical energy shock. Markets have not experienced this combination before. The combination of monetary uncertainty and geopolitical risk is novel, and the hearing is likely to produce more questions than answers — which is itself a form of volatility signal.
Live Nation Antitrust Verdict: A jury found that Live Nation and its Ticketmaster subsidiary hold an illegal monopoly over concert venues. Legal remedies under discussion include a potential structural break-up of the company. The verdict carries ESG governance implications for institutional holders of consumer discretionary indices and is unlikely to be resolved quickly.
IEA Demand Revision: The International Energy Agency trimmed its 2026 oil demand growth forecast by approximately 80,000 barrels per day. The fuel shock from the Hormuz blockade has already begun suppressing consumption. This is structural, not seasonal: demand destruction from an energy price shock takes months to reverse even after the price falls.
Science & Technology
Behind the Meter — the technology that is quietly rewiring how the world powers intelligence.
The Bloom Energy announcement this week — 2.8 gigawatts of solid oxide fuel cell systems committed to Oracle’s AI data centre network — was reported as a corporate deal. It is also a technology inflection point, and the distinction matters to anyone trying to understand where the AI power trade actually goes.
A solid oxide fuel cell (SOFC) generates electricity through an electrochemical reaction rather than combustion. Natural gas and oxygen combine across a ceramic membrane at temperatures above 700 degrees Celsius, producing electricity with an efficiency of roughly 60 percent — almost twice the efficiency of a conventional gas peaker plant and three times that of a diesel generator. The byproducts are water and carbon dioxide. There is no open flame. There is no turbine. There is very little that can go wrong in the traditional mechanical sense.
For AI data centres, the appeal is not primarily environmental. It is operational. A data centre running large language models at scale requires power that is uninterruptible, predictable, and immune to grid fluctuations. The grid cannot reliably offer all three simultaneously — not in California, not in Virginia, not in Texas, and not in most of the markets where hyperscale infrastructure is being deployed. Behind-the-meter generation — power produced on site, bypassing the transmission network entirely — solves all three problems simultaneously. Bloom’s systems can be sized from a single megawatt to hundreds, can be built in modular stacks, and can be operational in months rather than the years required for grid connection upgrades.
The 2.8-gigawatt Oracle commitment is the largest single SOFC deployment in history. To put it in scale: 2.8 gigawatts is the approximate generating capacity of three nuclear reactors, delivered in fuel cell stacks that fit inside standard industrial buildings. The capital cost per megawatt is still higher than grid power at the margin, but the cost trajectory for solid oxide technology follows the same curve that solar panels followed between 2010 and 2020. The question for investors is not whether behind-the-meter AI power is structural. This week answered that. The question is how quickly the cost curve compresses, and which companies own the manufacturing capacity when it does.
Bloom Energy is publicly listed and has been unprofitable for most of its existence as a public company. It became profitable in 2024. The Oracle deal is its largest customer announcement ever. The stock gained 23.98% on the news. The market is not wrong about the direction. Whether it is right about the distance is a different question.
There is a structural parallel in this week’s newsletter that is worth naming. The solid oxide fuel cell bypasses the grid bottleneck: instead of waiting for transmission upgrades, permitting delays, and the creaking logistics of centralised power delivery, it generates electricity on site, in months, at the point of need. This is precisely the logic that Neuralink applies to human cognition. The biological bottleneck — the hand, the eye, the processing speed of neural signals through conventional motor pathways — is bypassed at the source. In both cases, the insight is the same: the incumbent network is too slow, too expensive, and too unpredictable to serve the demands of the next decade. The smarter move is to route around it. The fuel cell does it for data centres. The neural interface does it for the human operator. The companies that own the bypass tend to define the architecture that follows.
Geopolitics
The distinction between open and resolved.
| Monitor | Status | Direction |
|---|---|---|
| 1. Hormuz / Middle East | Strait declared “completely open.” US naval blockade confirmed “in full force.” Ceasefire holds into Week 13. Note: the blockade’s legal standing under UNCLOS Article 38 (transit passage) is contested. Three major shipping insurers have activated force majeure clauses that survive a nominal “reopening” while the blockade continues. The February Supreme Court IEEPA ruling, which upheld executive authority to restrict maritime access on national security grounds, removed the principal legal constraint on indefinite enforcement. | → Cautiously De-escalating |
| 2. Fed Leadership Transition | Kevin Warsh confirmation hearing Tuesday 21 April, Senate Banking Committee. First mid-cycle Fed transition during an active energy shock in modern history. | ↑ New, Watch |
| 3. China Critical Minerals | Silver export limits (effective 1 Jan 2026) extend pattern of strategic resource leverage. Tungsten, antimony, refined silver now require government export approval. | ↑ Escalating |
| 4. US-China Trade | Tariff architecture unchanged. Legal ambiguity from February Supreme Court IEEPA ruling continues as background variable. | ↔ Stable |
| 5. EM Currency Stress | USD/TRY 44.85, -4.3% YTD. Turkey burns reserves defending the lira under energy import pressure — the central bank spent an estimated $24bn defending the currency during the last oil shock and has limited room for a repeat. USD/ZAR 16.31; rand +1.1% YTD, the only major EM currency in positive territory, benefitting directly from the metals supercycle. South Africa is effectively long copper and gold versus the world’s energy importers. The divergence is structural, not tactical. | ↑ Diverging |
| 6. UK-US Gilt Divergence | UK 10Y Gilt at 4.77% versus US 10Y at 4.26% — a 51 basis point spread that has widened steadily since February. The UK runs a larger energy import deficit as a share of GDP than the US. UK CPI is stickier: services inflation remains above 5%. The Bank of England is therefore constrained from cutting rates at the pace the Fed can, even if the Hormuz ceasefire proves durable. Sterling is also absorbing dollar strength more directly than the euro. The result is a gilt market that is pricing a more persistent inflation premium than equivalent US Treasuries — a divergence that matters to any portfolio that holds both. | ↑ Widening |
The currency table captures a question that the equity rally obscures entirely: if WTI holds above $80 through May, which emerging market central bank breaks first? Turkey is the most exposed. Its current account deficit widens approximately $3bn for every $10 per barrel increase in sustained oil prices. At $83 crude, that is a structural drain that reserve drawdowns and rate hikes can slow but not stop indefinitely. The rand tells the opposite story — a commodity currency in a world where the commodity story is copper and gold, not oil. These two currencies are not simply data points on a table. They are the clearest cross-asset signal of what this energy shock is actually doing to the global economy, country by country, balance sheet by balance sheet.
Signal vs. Noise
The number that does not fit the narrative — and why it matters more than the one that does.
This week’s anomaly: Michigan Consumer Sentiment 47.6 preliminary vs. S&P 500 at 7,126.
On the same Friday that the S&P 500 closed at a record 7,126, the University of Michigan released its preliminary consumer sentiment reading for April: 47.6. That is the lowest reading in the survey’s history. Lower than the 2008 financial crisis. Lower than the March 2020 pandemic shock. Lower than any point in the 1970s stagflation period.
One of these two numbers is noise. The question is which one.
The equity bull case: sentiment surveys measure how people feel, not what they do. Consumers have been persistently gloomier than their spending patterns suggest since 2021. The “vibecession” is real; the recession is not. The Hormuz reopening is structurally disinflationary, which is exactly what a rate-sensitive equity market wants to price in a single session. The record close reflects the relief, not the survey.
The bear case: 47.6 is not noise at these levels. Below 50, the University of Michigan survey has historically predicted a meaningful slowdown in consumer spending within two quarters in seven out of nine instances since 1978. Consumer spending is 68 percent of US GDP. The S&P 500 is trading at approximately 22 times forward earnings. At 22 times, forward earnings need to arrive. If consumer spending decelerates materially in Q3, they will not. The revised Michigan reading arrives next Friday. If it holds below 50, the equity record and the sentiment record will need to be reconciled by someone other than just the market.
Corporate profits and consumer sentiment have been running in opposite directions since 2022. The standard explanation — that sentiment surveys are capturing mood rather than behaviour — is partially correct but analytically incomplete. A more precise theory: corporate profits are being generated by a shrinking circle of beneficiaries. The S&P 500’s earnings growth in 2025–26 is overwhelmingly concentrated in the top decile of its constituents: the AI hyperscalers, the semiconductor complex, and the energy infrastructure names. These companies are genuinely profitable. Their employees are doing well. Their shareholders are doing extremely well.
The Michigan survey captures a different universe: the 160 million Americans whose income is not indexed to Nvidia or TSMC. For them, the energy shock is not a trading opportunity. It is higher petrol prices, higher utility bills, and a mortgage rate that has not fallen as fast as the Fed’s pivot implied it would. Sentiment at 47.6 is not a measurement error. It is a geographically and economically accurate reading of the half of the economy that the equity index does not represent.
This divergence can persist for longer than intuition suggests. But it cannot persist indefinitely. At some point, the consumer’s reality enters the corporate earnings cycle through the back door: weaker retail spending, weaker discretionary demand, weaker advertising revenue. The S&P at 7,126 is priced for a world in which the top circle keeps expanding. The sentiment survey suggests it is not.
Verdict: Noise in the short term. Structural signal in the medium term. The vibecession is not a measurement error — it is a distributional one. Watch the revised reading next Friday and the Q2 consumer discretionary guidance in May.
Contrarian Analysis
The consensus got the direction right. It may have got the duration wrong.
The consensus: This week’s record high is evidence the bull market is intact, the AI cycle is structural, and the worst of the geopolitical risk premium is behind us.
The contrarian case: A record high built in a single session, triggered by a single announcement, from a single foreign minister, while the naval blockade of the same waterway remains officially in place — this is a headline, not a thesis. The VIX at 17.48 tells you that professional hedgers have removed protection. History suggests that is precisely when protection becomes most expensive to replace.
Consider Netflix. The company reported its best quarterly earnings in years. The co-founder then announced his retirement. The second-quarter outlook disappointed modestly. The stock fell 9.72%. The market, in other words, punished Netflix for being good but not good enough, at the same moment it celebrated the S&P 500 for crossing a round number on a geopolitical headline. These two facts belong in the same paragraph.
The equity multiple currently embedded in 7,126 leaves limited room for guidance cuts. When guidance cuts arrive — and they will, somewhere, in some sector — the question is whether the market responds with the equanimity it showed on Friday or with the severity it showed Netflix. The answer depends almost entirely on whether the three assumptions in the Analytical Takeaway are still intact when the news arrives.
Perspective
One data point the week’s bad news is obscuring.
This week, global markets fretted about oil. A ceasefire in the Strait of Hormuz triggered one of the sharpest single-session crude price drops in years. The relief was immediate, visible, and loud. For a few days at least, the world’s most anxious chokepoint stepped back from the news.
But step away from the Strait for a moment and consider what has been happening to energy quietly, consistently, and almost entirely without fanfare for the past fifteen years.
In 2010, generating one megawatt-hour of electricity from solar panels cost approximately $380. Today, in most of the world, that figure is below $30. A 92% cost reduction in fifteen years. No energy source in recorded history has fallen in price that fast, for that long, without reversing.
The chart below shows the full arc. It is not a forecast. It is not a projection. It is what already happened, while the world was watching oil prices.
Source: Our World in Data / IRENA & Bloomberg NEF. Licence: CC BY 4.0. Interactive — hover to explore data points.
The Hormuz thread dominates the week’s tape because energy supply shocks are immediate and frightening. The solar transformation earns fewer headlines because it compounds slowly and happens in deserts, on rooftops, and in places that rarely feature on geopolitical risk maps. But the International Energy Agency’s own data shows renewables now accounting for the majority of new electricity capacity installed globally each year. The transition is not a prediction. It is already in the numbers.
The energy crisis the markets fear is real. The transformation that will eventually make such crises structural relics is also real, and considerably further along than this week’s headlines suggest.
The pessimists have better material. The data is on the other side.
Reading
Human Compatible — Stuart Russell’s argument that we are building something we do not yet know how to control.
Stuart Russell is one of the most cited computer scientists alive. His textbook on artificial intelligence has been a standard reference in the field for thirty years. In 2019, he published a book that was, in its way, a recantation: Human Compatible: Artificial Intelligence and the Problem of Control. The argument is simple to state and difficult to dismiss. We are building systems that are very good at achieving objectives. The problem is that we are not very good at specifying what those objectives should be.
The book’s central insight is what Russell calls the “King Midas problem.” Midas asked for everything he touched to turn to gold. He got exactly what he asked for. An AI system optimising for a specified objective will pursue that objective with a competence that humans cannot match — and will, in the process, find solutions that satisfy the letter of the specification while violating the spirit of it entirely. Russell’s argument is not that AI will become malevolent. It is that a system optimising for the wrong objective is, in practical terms, indistinguishable from malevolence in its effects.
This week’s newsletter put Neuralink’s thesis on the table: that the rational response to increasingly capable AI is to augment human intelligence rather than attempt to remain competitive against it. Russell would not necessarily disagree. But he would add a qualifier: the question is not just whether humans can keep pace with AI, but whether anyone — human or machine — is steering toward something worth reaching. A brain-computer interface that makes you faster at pursuing the wrong goal is still pursuing the wrong goal.
Human Compatible is not a doom narrative. It is an engineering argument dressed in accessible prose. Russell proposes a framework he calls “assistance games,” in which AI systems are trained not to maximise a fixed objective but to remain uncertain about human preferences and to defer to human correction. The insight is elegant: a machine that knows it might be wrong is safer than one that is certain it is right. The challenge is building the uncertainty in before the certainty takes over.
Human Compatible: Artificial Intelligence and the Problem of Control — Stuart Russell. Viking, 2019. 352 pages. Widely available in paperback.
And Finally
In which the news cycle reveals its priorities.
On Friday, as the S&P 500 crossed 7,100 for the first time in history, Netflix’s co-founder announced his retirement, Live Nation was declared an illegal monopoly, and the International Energy Agency quietly reduced its oil demand forecast. None of these received more than a footnote in the evening news. The Strait of Hormuz, technically, received all of it.
There is a principle at work in financial journalism: when an index hits a round number, other facts become optional. This week, the number was 7,000. Everything else — however consequential — was filed under “also.”
The Kevin Warsh confirmation hearing, described in pre-week briefings as the most market-sensitive Senate testimony since the Volcker era, is scheduled for 10am Tuesday. March Retail Sales data is released at exactly the same time. The Senate Banking Committee has thus managed to schedule the two most interest-rate-sensitive events of the month simultaneously. One assumes this was not deliberate. One is not entirely certain.
Buried in Friday’s celebration of 7,126 was a detail that deserves slightly more attention than it received: Reed Hastings, Netflix co-founder and the person most responsible for the company’s decision to move into original content a decade ago, announced his retirement from the board. Netflix stock fell 9.72%. The market, which had just cheered an index record, appeared to have a strong view on what the departure of a founding vision means for a streaming company in a crowded market. This is key person risk in its cleanest form. It is also a preview of a conversation that will become louder in the next 24 months, as the founders of the AI era’s defining companies approach the point where they become optional rather than irreplaceable. The question is not sentimental. It reprices.
A tariff truce. Seven-one-two-six.
A Warsh hearing. Some geopolitical tricks.
The VIX fell asleep,
The bulls counted sheep,
Conditionally bullish. That sticks.
Opening prices: 1 January 2026 locked baselines. Ranked by YTD performance.
| # | Asset | Class | 1 Jan ’26 | 17 Apr ’26 | YTD |
|---|---|---|---|---|---|
| 1 | Copper | Commodity (USD) | $4.07 | $6.10 | ▲ +50.0% |
| 2 | Baltic Dry | Commodity Index | 1,850 | 2,567 | ▲ +38.8% |
| 3 | Nasdaq 100 | Equity (USD) | 21,600 | 26,672 | ▲ +23.5% |
| 4 | WTI Crude | Commodity (USD) | $68.00 | $83.85 | ▲ +23.3% |
| 5 | DAX | Equity (EUR) | 20,073 | 24,702 | ▲ +23.1% |
| 6 | Euro Stoxx 50 | Equity (EUR) | 5,000 | 6,062 | ▲ +21.2% |
| 7 | Nikkei 225 | Equity (JPY) | 50,339 | 58,476 | ▲ +16.2% |
| 8 | MSCI EM | Equity (USD) | 1,397 | 1,604 | ▲ +14.8% |
| 9 | Swiss SMI | Equity (CHF) | 11,700 | 13,427 | ▲ +14.8% |
| 10 | Nifty 100 | Equity (INR) | 22,000 | 25,239 | ▲ +14.7% |
| 11 | Silver | Commodity (USD) | $71.74 | $80.75 | ▲ +12.6% |
| 12 | Gold | Commodity (USD) | $4,327 | $4,834 | ▲ +11.7% |
| 13 | Russell 2000 | Equity (USD) | 2,540 | 2,793 | ▲ +10.0% |
| 14 | AGG (US Agg) | Fixed Income | $94.00 | $103.12 | ▲ +9.7% |
| 15 | FTSE 100 | Equity (GBP) | 9,951 | 10,668 | ▲ +7.2% |
| 16 | TLT (Long Tsy) | Fixed Income | $90.00 | $95.12 | ▲ +5.7% |
| 17 | LQD (IG Corp) | Fixed Income | $105.00 | $110.04 | ▲ +4.8% |
| 18 | HYG (Hi Yield) | Fixed Income | $77.00 | $80.65 | ▲ +4.7% |
| 19 | S&P 500 | Equity (USD) | 6,858 | 7,126 | ▲ +3.9% |
| 20 | USD/ZAR | Currency | 16.49 | 16.31 | ▲ +1.1% |
| 21 | Hang Seng | Equity (HKD) | 26,361 | 26,160 | ▼ -0.8% |
| 22 | USD/TRY | Currency | 43.01 | 44.85 | ▼ -4.3% |
| 23 | Bitcoin (BTC) | Digital Asset | $87,725 | $75,864 | ▼ -13.5% |
| 24 | Ethereum (ETH) | Digital Asset | $2,968 | $2,349 | ▼ -20.8% |
| 25 | Natural Gas | Commodity (USD) | $3.50 | $2.67 | ▼ -23.6% |
| Index | Prior Close | Current | WoW Change | WoW % |
|---|---|---|---|---|
| S&P 500 | 6,783 | 7,126 | +343 | +4.54% |
| Nasdaq 100 | ~22,635 | 26,672 | +4,037 | est. +17.8% |
| Russell 2000 | 2,470 | 2,793 | +323 | +13.1% |
| FTSE 100 | 10,644 | 10,668 | +24 | +0.2% |
| Nikkei 225 | 55,895 | 58,476 | +2,581 | +4.6% |
| VIX | 19.33 | 17.48 | -1.85 | -9.6% |
| Instrument | Level | WoW Change |
|---|---|---|
| US 2Y Yield | 3.71% | est. -8bps |
| US 5Y Yield | 3.88% | est. -6bps |
| US 10Y Yield | 4.26% | -5bps |
| US 30Y Yield | 4.88% | est. +2bps |
| UK 10Y Gilt | 4.77% | — |
| German 10Y Bund | 3.03% | — |
| IG Credit Spread | 102bps | — |
| HY Credit Spread | 328bps | — |
| 10Y / 2Y Spread | +55bps | Steepening |
| Commodity | Price | WoW | Driver |
|---|---|---|---|
| WTI Crude | $83.85 | -11.45% | Hormuz “completely open”; CTA long liquidation |
| Brent Crude | $90.38 | -9.07% | June contract settled; war premium unwinding |
| Gold | $4,833.56 | +0.6% | Safe-haven bid persists; central bank diversification |
| Silver | $80.75 | est. +3% | China export limits; structural deficit; $100 target in view |
| Copper | $6.10/lb | +3.97% | AI data centre demand; tariff front-running |
| Natural Gas | $2.67/MMBtu | est. -4% | 17-month low; 108 Bcf above 5yr average inventory |
| Baltic Dry Index | 2,567 | — | Rerouting effects continuing; global shipping congestion |
| Asset | Price | WoW | Note |
|---|---|---|---|
| Bitcoin (BTC) | $75,864 | est. +4% | Retook $75k on Hormuz risk-on; “relief without reopening” |
| Ethereum (ETH) | $2,349 | est. +3% | Institutional accumulation pace double Bitcoin per QCP Capital |
| Stablecoin Mkt Cap | ~$190bn | — | — |
| Stock | WoW % | Catalyst |
|---|---|---|
| Bloom Energy (BE) | +23.98% | Oracle 2.8GW fuel cell deal; AI power infrastructure re-rating |
| Oracle (ORCL) | +14.0% | Bloom deal; AI utility product re-rating by JPM and Jefferies |
| Ally Financial (ALLY) | +8.10% | Q1 earnings beat; net interest margin stability |
| Royal Caribbean (RCL) | +7.34% | Oil -11% directly reduces largest variable cost |
| Caterpillar (CAT) | +2.85% | Strong industrial data; fresh 52-week high |
| Nike (NKE) | +2.81% | Recovery from 12-year lows; consumer discretionary optimism |
| Live Nation (LYV) | -6.30% | Jury finds illegal concert monopoly; break-up risk |
| Netflix (NFLX) | -9.72% | Co-founder retirement; Q2 guidance disappointed |
| Day | Event | Why It Matters |
|---|---|---|
| Mon 20 | PBOC Rate Decision | Any deviation from expectations triggers Asian tech and semiconductor volatility |
| Tue 21 | Kevin Warsh Confirmation Hearing (10am EST) | Rate path for 2026–27; hawkish signal compresses the equity multiple |
| Tue 21 | US Retail Sales (March) | Consumer durability after the fuel shock; simultaneous with Warsh |
| Wed 22 | UK CPI + EIA Crude Stock Report | UK inflation trajectory; storage depletion during the blockade |
| Thu 23 | S&P Global Flash PMIs (US, UK, Eurozone) | First comprehensive read on economic activity post-de-escalation |
| Fri 24 | Michigan Consumer Sentiment (revised) | Consensus 47.6 was record low preliminary; watch for rebound or deterioration |
| Monitor | Status | Direction |
|---|---|---|
| 1. Hormuz / Middle East | Strait “completely open.” US blockade “in full force.” Ceasefire Week 2. | → Cautiously De-escalating |
| 2. Fed Chair Transition | Warsh hearing Tue 21 Apr. First mid-cycle transition during active energy shock. | ↑ New, Watch |
| 3. China Critical Minerals | Silver, tungsten, antimony export limits active since 1 Jan 2026. | ↑ Escalating |
| 4. US-China Trade | Tariff architecture stable. IEEPA legal ambiguity ongoing. | ↔ Stable |
| 5. EM Stress | USD/TRY 44.85 (-4.3% YTD TRY). USD/ZAR 16.31 (+1.1% YTD ZAR). | ↑ Diverging |
| Strategy | Vehicle | Gross Yield | WoW | Risk |
|---|---|---|---|---|
| Active income + leverage | PIMIX | ~10.4% net | unch | Moderate |
| IG CLO (BBB) | Various | SOFR +192bps | +4bps | Moderate |
| Infrastructure debt | HICL, INPP | ~7.0% | unch | Lower |
| BDCs | FS KKR, ARCC | ~10.5% | -6bps | Moderate |
| Agency mortgage REIT | AGNC, NLY | ~13.8% | +10bps | Higher |
| Senior secured loans | BKLN | SOFR +395bps | +8bps | Moderate |
| Preferred shares | PFF | ~6.6% | +2bps | Moderate |
| Name | Role | Context This Edition |
|---|---|---|
| Elon Musk | Founder, Neuralink; CEO, Tesla & SpaceX | Entrepreneur section — Neuralink brain-computer interface |
| Abbas Araghchi | Iranian Foreign Minister | Declared Strait of Hormuz “completely open” — triggered Friday’s record rally |
| Kevin Warsh | Fed Chair nominee | Confirmation hearing Tuesday; most market-sensitive testimony of Q2 |
| Noland Arbaugh | First Neuralink human implant recipient | Entrepreneur section — paralysed patient who moved a cursor with thought |
| Anthony Rosenthal | Author, Weekly Market Pulse | Hero quote: “A ceasefire is not peace. It is simply the market’s permission to forget what frightened it.” |