Analysis

Has the World Bank Performed a U-turn on Industrial Policy? Interventionists Who Think So Should Read Its New Report More Closely

Published

3 months ago

April 24, 2026

The Bank’s landmark 2026 report is a significant intellectual evolution—but it is no blank cheque for state intervention. A careful reading reveals something more interesting, and more demanding, than either its cheerleaders or critics will admit.

When the Priest Revises the Catechism

When The Economist declared in April 2026 that the World Bank had abandoned three decades of stigma against industrial policy, the think-tank circuit lit up like a Christmas tree. Industrial policy advocates who had spent years being lectured about market distortions and government failure finally had what they thought was institutional absolution—from the very institution that had long served as the high church of the Washington Consensus. The Wall Street Journal, not typically given to rooting for state intervention, ran its own headline pronouncing that the World Bank had “embraced industrial policy.” The triumphalism from certain quarters of the development community was immediate, effusive, and—on closer inspection—substantially overblown.

The report in question, Industrial Policy for Development: Approaches in the 21st Century (March 2026), authored by economists Ana Margarida Fernandes and Tristan Reed, is a serious, carefully qualified, empirically grounded document that runs to several hundred pages of analysis drawn from 183 national development plans and evidence across more than 60 economies. It represents a genuine intellectual shift at the Bank—one worth examining in detail. But it is emphatically not the unconditional surrender to interventionism that its more excitable admirers have proclaimed. Those who are reading it that way are, to borrow a phrase, looking at a compass and claiming they’ve found a treasure map.

The Long Shadow of the Washington Consensus

To appreciate what has actually changed, it is necessary to recall what the old orthodoxy looked like—and how it came to feel so shopworn.

The Washington Consensus, the policy framework associated with John Williamson’s 1989 synthesis and subsequently operationalised by the World Bank and IMF across the developing world, was not a monolith of stupidity. It correctly identified the fiscal chaos, runaway inflation, and state capture that had ravaged Latin America and sub-Saharan Africa through the 1970s and 1980s. Privatisation, trade liberalisation, and macroeconomic stabilisation delivered genuine benefits in countries where the prior alternative had been kleptocratic mismanagement. To dismiss it entirely is intellectually dishonest.

But its treatment of industrial policy—the deliberate use of government instruments to shape the structure of an economy toward particular sectors, technologies, or firms—was always its weakest limb. The 1993 World Bank report, The East Asian Miracle, was compelled by the sheer empirical weight of South Korea, Taiwan, and Japan to concede that some forms of selective intervention had, in fact, accompanied extraordinary growth. Yet the report then executed what remains one of the more remarkable intellectual contortions in development economics: it simultaneously acknowledged that directed credit, export discipline, and sectoral targeting had been central to East Asia’s ascent, and concluded that this held “little promise” for most other countries. The reasoning—that East Asia’s state capacity was exceptional and unreplicable—was not without merit. But it served, conveniently, to leave the core doctrine of market supremacy largely intact.

That convenient wall has been crumbling for years. China’s state-led industrial rise, the CHIPS and Science Act in the United States, the European Union’s Green Deal Industrial Plan, the Inflation Reduction Act’s industrial subsidies—all represent major market economies abandoning the posture that selective state support for industries is inherently distortionary and therefore illegitimate. Against that backdrop, the World Bank clinging to the 1993 catechism would have rendered it not principled but simply irrelevant.

What the 2026 Report Actually Says—And What It Doesn’t

Indermit Gill, the Bank’s Chief Economist, frames the intellectual moment with admirable candour in his foreword. The 1993 report’s dismissal of selective industrial policy, he writes, has “the practical value of a floppy disk today.” It is a striking admission—frank to the point of self-deprecation—and it is why the headlines were understandable, if ultimately misleading.

Because when you move beyond Gill’s foreword and into the analytical body of the Fernandes-Reed report itself, what you find is not a celebration of state intervention but a sophisticated, heavily conditional framework for thinking about when and how industrial policy can work—and when it reliably fails.

Several findings deserve particular attention:

The tools are more diverse than the debate admits. The report catalogues 15 distinct policy instruments that governments deploy under the banner of industrial policy—ranging from performance-based subsidies and special economic zones to export promotion agencies, public procurement, and investment incentives. This taxonomy matters because much of the political debate treats industrial policy as synonymous with tariff walls and targeted subsidies. The Bank’s analysis suggests that the more successful contemporary interventions tend to operate through less blunt instruments: co-investment vehicles, matching grants conditional on export performance, and sector-specific infrastructure.

Upper-middle-income countries are already intervening heavily—and badly. One of the more arresting data points in the report is that upper-middle-income countries spend approximately 4.2% of GDP on business subsidies—a figure that rivals or exceeds what advanced economies deployed during the peak of post-war industrial planning. Developing economies, the report finds, are among the heaviest users of industrial policy instruments. The problem is not too little intervention; in many cases, it is poorly designed, poorly targeted, and poorly monitored intervention. This finding subtly reframes the policy debate: the question is not whether governments should engage in industrial policy but whether they should do it more intelligently.

Performance conditionality is non-negotiable. The Bank’s framework is insistent on what might be called the discipline condition. Effective industrial policy, the report argues, requires that support be time-bound, subject to measurable performance benchmarks, and genuinely withdrawable when those benchmarks are not met. The cautionary tale of subsidies that metastasise into permanent entitlements—zombifying industries rather than catalysing them—runs through the analysis as a recurring theme. This is not a departure from the Bank’s long-standing emphasis on institutional quality and accountability; it is a restatement of it in a new context.

Goals have multiplied beyond productivity. The 21st-century industrial policy toolkit, the report acknowledges, is being deployed in pursuit of objectives that would have seemed peripheral to the 1993 debate: job creation in specific regions, foreign exchange generation, green industrial transition, and national security resilience. The fusion of climate policy and industrial policy—manifest in the extraordinary state investments being made in clean energy supply chains across the US, Europe, China, and increasingly India—represents a structural shift in what governments are asking industrial policy to accomplish. The Bank’s framework attempts to provide analytical guidance across all these goals, though the tension between them is not always fully resolved.

Institutions still precede everything. For all the evolution in tone, the report is emphatic that the preconditions for successful industrial policy remain demanding. Strong bureaucratic capacity, credible commitment mechanisms, insulation from political capture, and a competitive domestic market environment are all listed as prerequisites rather than outcomes. This is where the interventionist reading tends to break down. The report is not telling governments with weak institutions, endemic corruption, and captured regulatory bodies that they should now feel liberated to pick winners. It is telling them, more carefully, that success under those conditions remains extremely unlikely—and that the sequencing question (fundamentals first) has not changed.

The Risks That Have Not Disappeared

None of the 20th century’s cautionary lessons about industrial policy have been repealed by the 2026 report. The risks of regulatory capture—where the industries being promoted come to shape the policies promoting them—remain as real as ever. The political economy of withdrawing support from failing industries has not become easier simply because the Bank has published a nuanced framework; it has, if anything, become harder in an era of economic nationalism where the political costs of being seen to abandon domestic producers are higher than ever.

The challenge of enforcement in low-capacity states deserves more attention than the report gives it. It is one thing to design performance conditionalities in theory; it is quite another to enforce them when the industry being supported employs 40,000 workers in a swing constituency, and when the monitoring agency lacks both the data systems and the political independence to apply sanctions. South Korea’s famous export discipline worked in part because the Park government was genuinely willing to withdraw credit from underperforming chaebol—a willingness that is historically unusual and politically contingent in ways that resist replication.

The report also underplays, perhaps intentionally, the geopolitical drivers of the current industrial policy revival. The CHIPS Act was not primarily a development economics exercise; it was a strategic response to China’s dominance of semiconductor supply chains and the perceived vulnerabilities that dependence exposed during the COVID-19 pandemic. The EU’s Critical Raw Materials Act is similarly animated by concerns about strategic autonomy that sit uncomfortably within a conventional welfare economics framework. When major powers justify industrial policy on national security grounds, they are not primarily inviting replication by developing countries—they are, in some respects, restructuring global supply chains in ways that create new dependencies for exactly those countries.

This is a significant gap. The World Bank’s mandate centres on development in the Global South, yet the industrial policy revolution currently reshaping global trade is being driven by the Global North for strategic reasons that may be actively harmful to developing country interests. A Bangladeshi garment manufacturer or a Kenyan software firm is not the primary beneficiary of the Inflation Reduction Act’s domestic content requirements; they may, in fact, be among its victims.

What the Report Gets Right

Sceptics who dismiss the 2026 report as ideological window-dressing—or as an institution capitulating to political fashion—are missing its genuine contributions.

The most important is evidentiary. The systematic review of 183 national development plans and the cross-country econometric evidence on policy effectiveness is the most comprehensive analytical exercise the Bank has conducted on this topic. It moves the debate beyond the anecdotal—beyond the duelling citations of Singapore’s success and Brazil’s Embraer against the failures of Tanzania’s groundnut scheme and India’s licence raj—and toward something more methodologically rigorous. The finding that well-designed export promotion agencies have positive effects on trade performance across diverse country contexts, for instance, is a useful practical contribution that deserves more attention than the headline debate about whether the Bank has “changed its mind.”

The 15-tool taxonomy is similarly valuable. It forces a more granular conversation. Blanket arguments for or against “industrial policy” obscure enormous variation in instrument design, targeting precision, conditionality structure, and institutional context. A matching grant for small manufacturing exporters in Vietnam is a fundamentally different policy animal from a permanent tariff wall protecting a state-owned steel company in Argentina, even if both travel under the same banner.

The report is also right to note that the conditions under which industrial policy operates have changed since 1993 in ways that are not purely political. Education levels and institutional baselines in many developing countries are substantially higher than they were 30 years ago. The technological infrastructure for monitoring and evaluation—the data systems, the satellite imagery for industrial zone oversight, the digital payment rails for conditional transfer programmes—has improved dramatically. The argument that East Asian-style industrial policy was uniquely unreplicable rested partly on state capacity arguments that are less universally true than they once were.

Implications for Developing Countries

For policymakers in developing economies, the 2026 report offers something more useful than either the old orthodoxy or the new triumphalism: a structured decision framework. The key questions it poses deserve wide circulation.

Which sectors or activities exhibit genuine market failures—information externalities, coordination problems, learning-by-doing spillovers—that justify intervention? Is the institutional capacity to design, monitor, and enforce conditionalities actually present? Are competition disciplines—from domestic rivalry or export markets—in place to prevent the support from degenerating into rent extraction? And is there a credible sunset mechanism, or is this a policy that will be permanent from the moment of its announcement?

These are demanding questions. They will not produce comfortable answers in many contexts. But they are the right questions—and the fact that the World Bank is now asking them openly, rather than simply proscribing the entire enterprise, is a genuine advance.

A Toolkit, Not a Theology

The appropriate metaphor for what the World Bank has done in March 2026 is not a U-turn. It is more like a careful renovation of a building that had become structurally unsound in certain sections while remaining sound in others. The macroeconomic fundamentals—fiscal discipline, monetary credibility, competitive exchange rates, strong property rights—remain in place as the ground floor. What the Bank has done is admit that the upper floors, specifically its prescriptions about the role of the state in shaping economic structure, need significant reconstruction.

Industrial policy, the 2026 report concludes, belongs in the development toolkit. But a toolkit is not an ideology. A skilled carpenter does not use a hammer for every job simply because a hammer is now considered acceptable; they use the tool that fits the problem, with the precision the job demands.

The interventionists celebrating a full reversal at the World Bank are indulging in the same binary thinking they correctly criticise in their opponents—they have simply flipped the polarity. The Bank’s new report is asking harder questions, not providing easier answers. For developing countries navigating a world of rising protectionism, accelerating automation, and green transition imperatives, that analytical discipline is precisely what is needed.

Whether governments will apply it with the rigour the Bank prescribes is, of course, an altogether different question. And it is the one that will determine whether the 21st century’s industrial policy renaissance looks more like South Korea in 1970 or Brazil in 1980. History suggests the answer will vary by country, by decade, and by the quality of the institutions doing the intervening. The World Bank has, to its credit, stopped pretending otherwise.

The market did not build the internet. It did not sequence the human genome. And it will not, on its own, decarbonise industrial civilisation on any timeline that matters. But governments that have failed to build functioning tax systems, independent judiciaries, and competitive markets are unlikely to succeed where markets have not. The World Bank’s new report understands this. The question is whether its readers do.

Further Reading and Sources:

Discover more from The Economy

Subscribe to get the latest posts sent to your email.

Up Next

Pakistan’s $3.45 Billion UAE Repayment: A Quiet Milestone in Debt Discipline or a Signal of Shifting Gulf Alliances?

Don't Miss

America’s Electoral Vandalism Crisis: Why Eroding Trust in Elections Threatens Democracy More Than Any Single Theft

Click to comment

Analysis

Strait of Hormuz 2026: Why Markets Still Don’t Trust It’s Open

Published

11 hours ago

July 11, 2026

Abdul Rahman

If you’ve followed headlines about the Strait of Hormuz over the past several months, you’d be forgiven for losing track of whether it’s actually open. That confusion isn’t a media failure — it genuinely has opened, closed, and reopened multiple times since the conflict began, and the pattern itself is the real story markets need to understand, far more than any single day’s price move.

A Timeline That Explains the Market’s Persistent Skepticism

The crisis began February 28, 2026, when US and Israeli military operations against Iran triggered Iranian retaliation, including drone, ballistic missile, and small-boat attacks on vessels attempting to transit the Strait (Brookings). By March 4, Iranian forces formally declared the Strait “closed.” Insurance for transiting vessels became unavailable or prohibitively expensive, and seafarers largely refused the journey — meaning the Strait was effectively shut even without a formal blockade in the technical sense (Brookings).

What followed was a genuinely chaotic sequence that explains why traders remain reluctant to fully price in a resolution even now. On April 9, there was no sign an earlier agreement to lift the blockade was actually being implemented — ships were once again prevented from passing. Abu Dhabi National Oil Company’s CEO confirmed the Strait remained closed despite an announced ceasefire, noting 230 loaded oil tankers were waiting inside the Gulf (Wikipedia — 2026 Strait of Hormuz crisis). On April 17, Iran’s foreign minister announced the Strait was open to all shipping — oil prices dropped 11% immediately following the announcement. The very next day, April 18, Iran closed it again, citing the US refusal to lift its own naval blockade in response.

Even the June 17 memorandum of understanding between Trump and Iranian President Masoud Pezeshkian to formally end the war and the blockades didn’t hold cleanly: on June 20, Iran said it had closed the Strait again, citing continued Israeli strikes in southern Lebanon as a violation of the broader ceasefire agreement — a claim the US military denied (Wikipedia). By June 27, the US Navy’s Joint Maritime Information Center announced a widened shipping route through the Strait near Oman, an action explicitly framed as challenging Iran’s control over the waterway rather than a clean bilateral resolution.

Why This Chokepoint Matters More Than Any Other Piece of Global Infrastructure

Approximately 20 million barrels of oil per day move through the Strait of Hormuz — roughly 20% of global seaborne oil trade and about 27% of the world’s maritime crude oil and petroleum product trade combined (Congressional Research Service). At its narrowest point, the Strait is just 33-34 kilometers wide, split into two unidirectional two-mile-wide shipping lanes separated by a two-mile buffer zone sitting entirely within Iranian and Omani territorial waters (Congressional Research Service).

Critically, no rerouting option exists that can replace this volume at comparable cost. An extended full closure would remove 17-21 million barrels from daily global supply against total world consumption of roughly 100 million barrels per day — a supply shock with no readily available substitute (Ziro Market).

The Damage Already Done, Even With Partial Reopening

The International Energy Agency characterized the disruption as the largest supply disruption in the history of the global oil market (Wikipedia — Economic impact of the 2026 Iran war). At peak conflict intensity in February-March 2026, Brent crude surged well above $120 per barrel. As ceasefire talks progressed through May and June, prices retreated significantly — falling to around $95-100 per barrel by early June, and briefly dipping to $78.24 per barrel by mid-June, the lowest level since March 3, before the framework agreement was formally signed (Al Jazeera).

But the ripple effects extend well beyond crude oil pricing. The Strait closure disrupted roughly 45% of global sulfur supply — critical for fertilizer production, copper industry metal leaching, and sulfuric acid manufacturing — and constrained helium supply, a commodity essential to semiconductor manufacturing (Wikipedia — Economic impact). Shipping companies including Maersk, CMA CGM, and Hapag-Lloyd suspended transits through the Strait and related routes like the Red Sea entirely, forcing rerouting around the Cape of Good Hope that added two to three weeks to journey times and increased per-shipment costs by 30-50% (Ziro Market).

Europe’s Quieter But Deeper Crisis

While oil price headlines dominated coverage, Europe faced an arguably more severe parallel crisis through the suspension of Qatari liquefied natural gas exports combined with the Strait closure — hitting at the worst possible moment, with European gas storage sitting at just 30% capacity following a harsh 2025-2026 winter. Dutch TTF gas benchmarks nearly doubled to over €60/MWh by mid-March (Wikipedia — Economic impact).

The European Central Bank responded by postponing planned interest rate reductions on March 19, simultaneously raising its 2026 inflation forecast and cutting GDP growth projections, with UK inflation specifically projected to breach 5% during 2026. Chemical and steel manufacturers across the UK and EU imposed surcharges of up to 30% to offset surging electricity costs, and the ECB explicitly warned that a prolonged conflict risked pushing major energy-dependent economies, including Germany and Italy, into technical recession by year-end.

Why OPEC+ Couldn’t Simply Fill the Gap

A natural question is why Saudi Arabia and the UAE — the two largest Gulf Cooperation Council producers with meaningful spare capacity — didn’t simply increase output to compensate. The answer is logistical rather than a lack of willingness: the Strait closure itself limited their ability to actually export any increased production volumes, even when pumping more oil, because the export bottleneck was the same chokepoint causing the broader crisis (Ziro Market). Total OPEC country production fell more than 30% since the start of the war, and the region’s spare capacity — the traditional shock absorber for global oil markets — proved largely irrelevant when the actual export route itself was under attack (Brookings).

US shale producers, meanwhile, responded more slowly to the price signal than historical patterns would predict. Rig counts stayed largely steady through April 2026, though well-completion activity in the Permian Basin did rise roughly 20% over several weeks as previously drilled wells came into production — still below pre-pandemic activity levels overall (Brookings).

The Market Is Still Pricing a Discount for Uncertainty, and Analysts Say That’s Correct

Vandana Hari, founder of Singapore-based Vanda Insights, offered perhaps the most useful framing for understanding current market behavior: crude’s slide following the memorandum of understanding is “entirely sentiment-driven,” with markets front-running the prospective reopening and likely pricing in a best-case scenario for normalized flows — meaning potential hiccups, from logistics to renewed geopolitical tensions, aren’t being adequately factored in (Al Jazeera).

Given the actual track record — multiple announced reopenings followed by renewed closures throughout April and June — that skepticism looks well-founded rather than excessive.

What This Means for Businesses and Investors Going Forward

For companies with Gulf-dependent supply chains: Treat any single reopening announcement as provisional rather than a genuine all-clear, given the pattern of reversals throughout the spring. Maintaining rerouting contingency plans and insurance flexibility remains prudent even after formal ceasefire signings.

For inflation-sensitive investors and central bank watchers: The relationship Ziro Market’s analysis highlights is worth internalizing directly: whether oil settles near $80-85 (supporting rate cuts, lower CPI, stronger oil-importing currencies) or spikes back toward $120 (elevated inflation, delayed rate cuts) functions as a genuine macro regime switch — not a marginal input, but potentially the single largest swing factor for 2026 global monetary policy.

For commodity-exposed sectors beyond energy: The sulfur, fertilizer, and helium supply disruptions are underappreciated second-order effects that specifically hit agriculture and semiconductor manufacturing — sectors not typically associated with Middle East conflict risk but directly exposed through this specific chokepoint.

The Bottom Line

The Strait of Hormuz crisis of 2026 has been less a single supply shock than a recurring pattern of partial resolutions and renewed disruptions, and that pattern itself is the most important thing for markets and businesses to understand going forward. Prices have retreated substantially from their conflict-peak highs, and the June 17 memorandum of understanding represents genuine diplomatic progress. But given that the Strait has been declared “open” and then closed again multiple times within the same several-week windows, treating the current relative calm as a durable resolution — rather than the latest phase in an ongoing negotiation — would be a mistake that both markets and policymakers seem determined not to repeat.

Discover more from The Economy

Subscribe to get the latest posts sent to your email.

AI

AI Capex Bubble 2026: The Hidden $662B Debt Nobody Reports

Published

11 hours ago

July 11, 2026

Abdul Rahman

Every earnings season now brings a fresh wave of headlines about hyperscaler AI capital expenditure hitting a new record. The “big four” — Amazon, Microsoft, Alphabet, and Meta — are on track to spend roughly $725 billion combined in 2026, a 77% jump from the $410 billion deployed in 2025 (UnboxFuture). That number gets reported constantly. What almost nobody is reporting with the same prominence is a separate figure that may matter more: roughly $662 billion in data center lease commitments that hyperscalers have already signed but not yet begun — obligations that currently sit entirely off balance sheet.

Why the Off-Balance-Sheet Number Changes the Whole Picture

Under GAAP accounting rules governing when a lease “commences,” these signed-but-not-started commitments don’t appear in the capital expenditure figures analysts and investors typically scrutinize when assessing hyperscaler financial health. According to reporting citing Moody’s early-2026 analysis, this shadow liability is larger than the combined on-balance-sheet debt of the same companies (Anomaly Investments).

That detail matters enormously for one specific argument AI infrastructure bulls have relied on: the claim that this buildout is being conservatively self-funded from operating cash flow rather than risky leverage. Once the full picture of committed-but-unrecognized obligations is accounted for, that defense becomes much harder to sustain.

The Debt Is Already Showing Up, Not Just Theoretical

This isn’t a purely hypothetical concern about future liabilities. Big tech companies have already issued more than $100 billion of bonds in 2026 specifically to help fund AI capital expenditure, and investors have responded by demanding record levels of protection against potential defaults through credit default swaps — essentially insurance policies against bond default (IEEE ComSoc).

Individual company examples illustrate the shift toward leverage: Oracle issued an $18 billion bond specifically tied to its data center expansion; CoreWeave secured a $2.6 billion loan alongside a $1.75 billion bond package; and OpenAI and Oracle reportedly entered into a $100 billion vendor financing arrangement (Anomaly Investments). At Amazon specifically, capital expenditure over the trailing twelve months has reached $151 billion — a figure that now exceeds the company’s entire operating cash flow, pushing free cash flow into negative territory.

The Depreciation Assumption Almost No Coverage Questions

Here’s an angle genuinely underexplored across most financial media: the depreciation schedules hyperscalers use for AI hardware assume a five-to-six-year useful life. But given how rapidly GPU generations are turning over and how intensively AI workloads are pushing hardware utilization, critics argue the real economic life of this equipment is closer to two to three years. That gap between assumed and actual depreciation is estimated to understate true asset depletion by roughly $176 billion between 2026 and 2028 alone — a figure that grows as accelerating token consumption pushes hardware utilization beyond the assumptions built into current depreciation schedules (Anomaly Investments).

Layered on top of that is the energy cost curve: running the current roughly 30-gigawatt installed base of AI infrastructure costs approximately $27 billion annually today, but that figure is projected to climb to between $45 and $90 billion per year as capacity scales toward 2029 — and crucially, these are first charges against revenue, not optional or deferrable costs.

The Revenue Gap: Who’s Actually Paying for All This?

The most commonly cited justification for the capex surge is that the pure-play AI vendors — OpenAI, Anthropic, and others — represent a massive and rapidly growing revenue opportunity. The reality is more nuanced. OpenAI’s roughly $20 billion annualized revenue run rate, while genuinely impressive for a company with barely any consumer products three years ago, represents only about 3% of projected 2026 hyperscaler capex. Anthropic’s roughly $9 billion run rate, despite showing 9x year-over-year growth, occupies a similarly small share. The entire cohort of pure-play AI vendors combined — including Cohere, Mistral, Perplexity, and others — likely accounts for less than $35 billion in projected combined 2026 revenue against a hyperscaler capex figure exceeding $700 billion (Futurum Group).

That gap is the crux of the bubble debate: hyperscalers are betting the infrastructure will ultimately serve enterprise adoption and their own AI services broadly, not just third-party AI vendor revenue — but that bet requires enterprise AI monetization to arrive at a scale that, as of mid-2026, remains largely unproven outside of code generation and basic customer service automation.

The Skeptic’s Case, From Inside Goldman Sachs Itself

The most prominent voice of institutional skepticism doesn’t come from an outside critic — it comes from within Goldman Sachs itself. Jim Covello, the bank’s Head of Global Equity Research, has consistently argued the economics of the generative AI transition are fundamentally flawed, stating in mid-2026 that the industry has moved “further away” from justifying the scale of capital expenditure compared to two years prior (UnboxFuture). Covello has specifically flagged circular capital flows between cloud providers and AI startups — where hyperscalers invest in AI companies that then spend that same capital purchasing compute from those same hyperscalers — as a red flag reminiscent of vendor financing patterns seen in the dot-com era.

The valuation comparison to that era is explicit and increasingly common among strategists: US technology and AI equities carry EV/EBITDA multiples near 25x, close to historical extremes and above the telecom valuations that preceded the 2000 dot-com peak. More specifically, capex is currently expanding roughly 46 percentage points faster than revenue growth — a gap that exceeds the 32-point divergence observed during the 2001 telecom excess cycle (Allianz Research). Separately, Bank of America strategists have pointed out that AI stock concentration has reached levels matching prior bubble peaks, with the “AI Big 10” (Nvidia, Microsoft, Alphabet, Amazon, Meta, Apple, Tesla, Broadcom, Micron, and AMD) now making up 41% of the S&P 500 — comparable to the concentration of tech and telecom stocks during the actual dot-com bubble (Yahoo Finance).

The Bull Case Isn’t Naive Either

It would be inaccurate to frame this purely as informed skeptics versus blind enthusiasm. Goldman Sachs’ own broader research (distinct from Covello’s individual view) models roughly $7.6 trillion in cumulative AI capital expenditure between 2026 and 2031, built on the expectation that token consumption will increase 24-fold by 2030, driven largely by enterprise AI agents becoming embedded in production workflows rather than remaining experimental (Sesame Disk / Goldman commentary). Microsoft has disclosed an $80 billion backlog of Azure orders it currently cannot fulfill due to power constraints — genuine evidence that demand, at least for existing capacity, is outpacing even the current aggressive build-out pace (Futurum Group).

Leverage levels also remain more conservative than headlines suggest in absolute terms: the top five US capex providers reported a combined $385 billion in debt at the end of 2025, with leverage ratios still roughly 20% below the “high spender” cohort from the 2000 dot-com peak, according to Allianz Research analysis — meaning rising debt levels are a trend worth monitoring closely, not yet an acute crisis.

What Happens If the Bubble Skeptics Are Right

Historical infrastructure cycles offer a specific and somewhat counterintuitive lesson: the investors who fund the initial frenzied build-out phase rarely capture the long-term rewards. If the AI capex cycle follows the pattern of the 1998-2001 fiber optic buildout, hyperscalers may eventually be forced to write down the value of data centers and GPUs purchased at today’s prices and utilization assumptions. But that collapse in computing costs, paradoxically, could pave the way for a new generation of leaner, genuinely profitable software companies to build on top of the resulting cheap, overbuilt infrastructure — much as fiber-optic overbuild eventually enabled the 2000s streaming and cloud computing boom, even after the original telecom investors were wiped out.

What This Means for Investors and Businesses

For equity investors, the practical signal to watch isn’t the headline capex number — it’s the widening gap between capex growth and revenue growth, and whether that gap begins narrowing through 2027 as enterprise adoption either accelerates or disappoints. For businesses evaluating AI vendor relationships, the circular-financing pattern flagged by Covello is worth diligence: understanding whether an AI vendor’s revenue depends partly on capital originally supplied by the same hyperscaler providing its compute is a legitimate red flag for assessing that vendor’s underlying financial independence. For fixed-income investors, the rising credit default swap pricing on hyperscaler-linked debt is itself a market signal worth tracking as an early indicator of shifting sentiment, independent of equity price action.

The Bottom Line

The AI infrastructure buildout genuinely is the largest corporate capital expenditure cycle in recorded history, and it’s happening for real, defensible reasons tied to a genuine technology shift. But the debate over whether it constitutes a bubble isn’t really about whether AI technology is useful — it’s about whether the timing of returns can keep pace with public equity markets’ patience, and whether the $662 billion in off-balance-sheet lease commitments, aggressive depreciation assumptions, and circular vendor financing arrangements represent manageable financial engineering or the early architecture of a genuinely serious correction. Both cases have real evidence behind them. What’s clear is that the headline capex figure everyone quotes is no longer the most important number in this story.

Discover more from The Economy

Subscribe to get the latest posts sent to your email.

Markets & Finance

Gold Overtakes US Treasuries in Reserves: What It Means

Published

11 hours ago

July 11, 2026

Abdul Rahman

Most gold coverage in 2026 has fixated on the price chart — the spectacular run from roughly $2,633 an ounce at the start of the year to fresh record highs above $5,400 by mid-year (Intellectia). That’s a legitimate story. But it’s not the most important one. The more consequential shift is structural, not seasonal: gold has overtaken US Treasuries as the largest share of global central bank reserves for the first time in three decades (BlackRock).

That’s not a headline about a commodity rally. It’s a headline about the architecture of the global monetary system quietly shifting under everyone’s feet.

The Trigger Most Coverage Undersells

The pivotal moment behind this shift traces back to 2022, when roughly $300 billion of Russian central bank foreign exchange reserves were frozen as part of international sanctions following the invasion of Ukraine (ISA Bullion). For reserve managers around the world — not just in Russia — that event functioned as a wake-up call: dollar-denominated assets held abroad are not unconditionally safe from geopolitical sanctions risk. Gold, by contrast, carries no counterparty risk; nobody can freeze a gold bar sitting in a country’s own vault.

That single realization has reshaped reserve management strategy globally. Central bank gold purchases averaged 225 tonnes per quarter between 2021 and 2025 — roughly double the pace seen from 2016 to 2020 (J.P. Morgan Global Research). BRICS+ nations now hold 17.4% of global gold reserves, up sharply from just 11.2% in 2019 (ISA Bullion).

Who’s Actually Buying, and Why the List Matters

Poland has been the standout accumulator, adding 20.2 tonnes in February 2026 alone, another 11.2 tonnes in March, and 14 tonnes in April — extending a rapid buildup that has added more than 360 tonnes to its reserves since 2023 (BestBrokers). China’s central bank maintained consecutive monthly gold purchases for 19 straight months through May 2026, even though much of this buying goes officially unreported to the IMF — analysts widely believe the People’s Bank of China continues accumulating gold “off the books” (ISA Bullion).

China’s motivation appears explicitly strategic rather than opportunistic. Chinese net gold imports jumped to 317 tonnes in the first quarter of 2026 alone — nearly triple the prior quarter — while the People’s Bank of China’s own reported purchases accelerated from roughly one tonne per month through February to eight tonnes in April (J.P. Morgan Global Research). J.P. Morgan’s own analysts frame this as part of a long-term Chinese project to build gold reserves as a foundation for establishing the renminbi as a credible alternative reserve currency.

A World Gold Council survey found a striking 95% of central banks expect to increase their gold holdings in 2026, up from 81% in 2024 and just 52% in 2021 — a trajectory showing accelerating, not plateauing, institutional conviction (BlackRock).

The Part of the Story Most Coverage Misses: Not Everyone Is Buying

Here’s an angle that gets consistently underplayed: this isn’t a uniform global stampede into gold. Several countries, including Singapore, Jordan, Mexico, and the Solomon Islands, actually reduced their gold reserves in 2025 — Singapore in particular emerged as a notable seller, likely driven by portfolio rebalancing decisions and a desire to realize gains after gold’s historic surge, rather than any lack of confidence in the metal (BestBrokers). Germany, for its part, has reduced its gold holdings every year since at least 2002, though its 2024 sale of just 1.1 tonnes was the smallest annual reduction on record.

This nuance matters for anyone trying to build a genuinely accurate picture: the de-dollarization and gold-accumulation trend is heavily concentrated among specific emerging-market and non-aligned economies — not a universal central bank consensus. Understanding which countries are buying and why is more analytically useful than simply citing an aggregate global purchasing figure.

Where Forecasts Diverge — And Why the Spread Is So Wide

Institutional price forecasts for gold currently show a genuinely unusual spread. J.P. Morgan projects gold reaching $6,000 an ounce by the end of 2026, and potentially $6,300 by the end of 2027 (J.P. Morgan Global Research). Morgan Stanley’s more conservative 2026 forecast sits at $4,400 an ounce (Morgan Stanley), while State Street projects a range of $4,750 to $5,500, and DWS targets $5,400 by mid-2027 (Discovery Alert).

A spread exceeding $1,500 per ounce between the most bullish and most conservative institutional forecasts reflects a genuine, unresolved analytical disagreement — not just differing house styles. The bull case rests on the idea that central bank reserve diversification represents a structural, policy-level shift rather than opportunistic market timing, making it fundamentally different from prior gold cycles driven mainly by retail or momentum investors. The more cautious case notes that gold’s roughly 245% rally from September 2022 to January 2026 is the largest percentage advance in modern gold market history — and historically, rallies of that magnitude have eventually triggered significant, multi-year corrections (Discovery Alert).

The Under-Discussed New Buyer: Stablecoin Issuers

One of the least-covered developments in this entire gold story is the emergence of stablecoin issuers as a genuinely new category of gold demand. As crypto markets have matured, some stablecoin issuers have begun holding gold as part of their reserve backing strategy — a development BlackRock specifically flags as part of the “early stages” of a new demand wave that also includes central banks and the broader AI infrastructure buildout’s effect on institutional portfolio hedging behavior (BlackRock).

What This Means for Different Audiences

For everyday investors: Gold ETPs still make up only about 0.17% of total US private financial assets, remaining well below prior peaks seen in the early 2010s, while private wealth gold allocations globally sit roughly 50% below levels seen a decade ago (BlackRock). That suggests meaningful room for incremental Western retail and institutional demand to grow, even after the current rally, if the structural de-dollarization narrative continues to gain mainstream acceptance.

For businesses managing currency exposure: The scale and persistence of central bank gold buying is one of several signals (alongside Fed communication policy changes and fiscal deficit concerns) suggesting continued structural pressure on the US dollar’s long-term reserve currency dominance — a trend worth factoring into multi-year currency hedging strategies rather than treating as a short-term news cycle.

For portfolio allocators: The unusually wide spread between institutional forecasts is itself useful information — it suggests treating any single gold price target as a scenario input rather than a confident base case, and sizing gold allocations based on its role as a portfolio diversifier and inflation/geopolitical hedge rather than as a directional price bet.

The Bottom Line

The gold price chart is the story most people are watching. The reserve-composition shift is the story that actually matters for the long-term structure of global finance. Gold surpassing US Treasuries as the largest share of central bank reserves for the first time since 1996 is a genuinely historic threshold — one triggered specifically by the 2022 Russian asset freeze and now sustained by a broad, if uneven, cohort of emerging-market central banks pursuing deliberate de-dollarization strategies. Whether the price keeps climbing toward J.P. Morgan’s $6,000 target or cools toward Morgan Stanley’s more conservative range matters less, in the long run, than the structural fact that the world’s reserve managers have permanently changed how they think about gold’s role in the global financial system.