TheGiltBook — The Record

The Big Picture · Macro & Policy Trends

Donald Trump flew to Beijing expecting a deal. What Xi Jinping offered him was a dilemma — which is a different thing entirely, and considerably more expensive.

The summit communiqué delivered one of the three phrases The Record identified as the operative test: "strategic stability framework" arrived on schedule, anchoring a three-year bilateral architecture in which cooperation is the mainstay, competition stays within "proper limits," and differences remain — the communiqué's word — manageable. "Freedom of navigation" did not appear. "Sanctions relief pathway" did not appear. The Hormuz question emerged as a concern about "global energy security and supply chain stabilisation" — the diplomatic register for: we talked about it and agreed it was a problem.

Xi's sharpest language was reserved not for Iran but for Taiwan, which he called the "most important issue in China-US relations" and warned could produce "direct collision" if handled poorly. The message to the Trump delegation — which included Elon Musk and Jensen Huang, whom Xi welcomed to expand operations in China — was precise: commerce is available; the security architecture is not negotiable; and the price of stability is restraint on the issues that matter to Beijing.

Trump declared "fantastic trade deals" on Air Force One. Analysts parsing the readouts found a trade truce extended, rare-earth export restrictions paused, and communication channels reopened. These are not nothing. They are also not Hormuz.

The market had priced the meeting. It is now pricing what the meeting produced. The Strait remains closed. The oil premium remains in the price. The Bayesian gap identified in Issue 18 — hope at 55%, evidence at 17% — has not closed. It has been given a diplomatic frame and called strategic stability.

Trump faces the horns of a dilemma. Accept Beijing's architecture and cede the post-1973 energy order quietly; or reject it and own the oil price, the S&P, and the political cost at home. Taiwan has sharpened the geometry. Xi opened the summit with an explicit warning about "conflicts" over the island; Trump left Beijing unsure whether to proceed with a $14 billion arms package that Xi had warned against in February. "The last thing we need right now is a war that's 9,500 miles away," Trump told reporters on Air Force One — a sentence that will have been read in Taipei, Tokyo, and Seoul with something considerably colder than relief. What the Beijing summit established is that Xi understands Trump's dilemma perfectly — on oil, on Taiwan, on the architecture of the post-war order — and is in no hurry.

United Kingdom

The UK's public relations have always been worse than its balance sheet. While Westminster provides the daily theatre and the bond market provides the tantrums, the underlying structural position is considerably more defensible than the yield on the 10-year Gilt currently implies.

The fiscal fundamentals are genuinely misread. The UK carries the second-lowest public debt-to-GDP ratio in the G7 — a fact that sits awkwardly alongside a borrowing cost that prices it as a serial offender. Combined public and private debt at 219% of GDP makes the broader economy more resilient to prolonged shocks than its highly leveraged competitors. The current yield premium is a political risk surcharge, not an economic one. Politics is cyclical; balance sheets are structural.

The productivity picture is quietly improving, an annualised rate of 1.6 per cent, which compares with less than 0.3 percent per year in the previous decade. The UK is the third-largest AI market globally, behind only the US and China, and is approaching the inflection point of the technology J-curve — the phase where experimentation becomes integration and structural growth accelerates. If that transition lifts long-term growth to 2–3%, it rewrites the arithmetic on public finances entirely.

The energy position is turning. High electricity prices have battered the industrial base for years; that period is ending. The UK is converging toward, or undercutting, German and Japanese energy costs, and is on track to become a net exporter of clean power by the early 2030s — removing a structural vulnerability that markets have long treated as permanent.

The Gilt yield at 5.18% reflects the political noise. The structural case suggests the noise is temporary. The fundamentals are not.

Stock Markets

The summit trade — what was priced, what was not. The S&P 500 at 7,408 and the Nasdaq at 26,225 represent a market that has absorbed the Beijing communiqué without a reckoning. The relief rally that preceded the summit has been partially sustained; the disappointment that the summit warranted has been partially deferred. This is not unusual. Markets price narratives faster than they price outcomes, and the narrative — a meeting, a framework, a handshake — remains intact even when the outcome does not support it. The oil price is the corrective mechanism. WTI at $102.12 and Brent at $106.64 are not consistent with the equity valuations currently on the screen. One of them will move toward the other. History suggests it is rarely the oil price that blinks first.

Friday's session ended with a broad sell-off — not a crash, but the kind of orderly retreat that carries more information than a panic: investors reducing risk into the weekend with clarity about what the summit did and did not deliver.

The AI capital cycle — the structural argument markets have not yet priced. The five hyperscalers have committed between $630 billion and $725 billion in capital expenditure for 2026 alone. The question those commitments raise is not whether AI is a real technology — it is — but whether the returns on the capital being deployed will justify the investment. The historical answer, applied to every comparable technology cycle, is uncomfortable.

If AI ultimately fulfils the bull-case promise of becoming genuinely transformative, the primary beneficiaries will be the users, not the providers. AI will be cheap, abundant, and widely available, with enormous productivity benefits accruing to enterprises and individuals. But cheap, abundant, and widely available is the opposite of pricing power. It is also the historical echo of an industry that creates large consumer surplus and modest producer profit. The hyperscalers, the model labs, and the silicon makers are funding the build-out; the customers will reap the benefits. This is a feature of how transformative technologies typically work, and it is the principal reason the equity returns to the capital deployed in this cycle are likely to disappoint.

The self-reinforcing mechanics are already visible. Power scarcity raises capex; raised capex requires more capital; tighter capital pushes boards toward discipline; discipline cuts the demand for memory and silicon; lower demand pulls forward the margin compression at NVIDIA, Broadcom, and the HBM oligopoly. The system is interconnected in either direction — and it is currently running in one direction only.

The historical parallels are precise. Rail in 1893, telecom in 2000–2002, shale in 2014–2020, and now AI in 2026: capital pursues a real technology revolution faster than the technology can monetise; capacity is built; consolidation phases follow; the ultimate winners are often companies that did not exist or were small at the cycle peak. The economic value of AI is real and will be enormous over decades. The investor return on the capital being committed at this stage of the cycle is the question, and history is not on the bulls' side. The most likely moment of recognition — the "aha" — arrives not on a geopolitical headline but in an earnings season. Q2 reporting begins in July.

Gold	4,530	MACD bearish signal — safe-haven bid fading on summit optics
Copper	13,530	Friday price reversal — demand outlook uncertain
WTI	$102.12	Back above $100 — Hormuz premium intact; no resolution
Brent	$106.64	Spread widening — physical supply constraints confirmed
Carbon	75.75	MACD neutral — correlated to gas; watching energy settlement

UST 10Y	4.60%	2Y at 4.08% — yield moving higher; oil and fiscal pressure; no Fed cut in sight
UK Gilts	5.18%	New cycle high — political and oil pressure compounding
Bund 10Y	3.18%	Rising — defence fiscal expansion; supply pressure ongoing
JGB 10Y	2.72%	Continuing to rise — BoJ normalisation; carry trade under pressure

Market Opportunities & Fears

The Fears

— The summit was the exit, not the entry. The S&P at 7,408 with WTI at $102 and Gilts at 5.18% is not a stable configuration. The Beijing communiqué has extended the rally's life by days, not weeks. The oil price has not moved toward the equity price; the equity price has not moved toward the oil price. Something will give, and the mechanism is Q2 earnings — beginning in July — where the Hormuz impact on corporate revenues arrives in reported numbers rather than analyst estimates. Expected market reaction: Equity indices hold near highs into June; the adjustment is concentrated and disorderly in earnings season. Historic memory points to a reduction in equity risk before the reporting window opens, not during it.

— The AI capital cycle turns visible in Q2. The hyperscaler capex commitments of $630–725 billion for 2026 are not in question. The returns on that capital are. When cheap, abundant AI becomes the market's consensus — which the technology's own trajectory makes inevitable — pricing power collapses and consumer surplus expands. The equity market appears not to have priced this transition. NVIDIA, Broadcom, and the HBM oligopoly are the most exposed when the margin compression cycle begins. Historic examples — Rail, telecom, shale: the pattern is consistent. The ultimate winners are rarely the companies that were largest at the cycle peak. Expected market reaction: No single trigger — the repricing is gradual until it is sudden. Q2 earnings guidance is the first moment management teams must quantify the return on deployed capital. Watch the language, not just the numbers.

— Taiwan — the price that has not yet been invoiced. Trump's uncertainty over the $14 billion arms package is not a negotiating position; it is a structural shift in the US security guarantee that has underwritten Asian equity valuations for decades. Takaichi's hawkish line is now explicitly out of step with Washington. South Korean and Japanese assets are priced for a security architecture that may no longer exist in its previous form. Expected market reaction: Asian equity risk premiums are structurally mispriced. The adjustment is slow until a specific event accelerates it — an arms sale decision, a Taiwan Strait incident, or a Trump statement that removes remaining ambiguity.

The Opportunities

— Gilts at 5.18% — a historically significant yield level. The argument made in Issue 18 at 5.00% is arithmetically stronger at 5.18%. For a UK investor in the higher-rate tax bracket, a 10-year Gilt at this yield delivers a pre-tax return that, on a risk-adjusted basis, has rarely been available outside periods of acute crisis. The political noise, the oil pressure, and the Bank of England's current paralysis are the mechanism that has created the yield level — they are not, in themselves, the reason to hold. The structural thesis remains: if oil falls and the rate cycle resumes, Gilt duration produces capital gains for those who held through the noise. That sequence has not yet begun. Whether and when it does is a matter of judgement for each investor and their adviser.

— Duration globally — what the history of rate cycles suggests. UST 10Y at 4.60%, Bunds at 3.18%, JGBs at 2.72%: every major sovereign bond market is pricing a persistent inflation and supply shock. Historically, when growth fears overtake inflation fears — a transition that earnings seasons tend to accelerate — longer-duration government bonds have been among the principal beneficiaries. The AI capital cycle argument, if the historical parallels hold, suggests that transition arrives in Q2 reporting. Whether that pattern repeats is not certain; what is observable is that the current yield levels reflect a set of assumptions about growth and inflation that Q2 earnings will either confirm or complicate.