Showing posts with label Commodity Market. Show all posts
Showing posts with label Commodity Market. Show all posts

Friday, November 20, 2026

Compare AI debt to dot-com bubble.

AI debt differs from the dot-com bubble primarily in financing structure and scale of physical infrastructure, even as both involve over-optimism about a transformative technology. The ECB and others explicitly separate the comparisons: the scale of AI capital spending resembles (or exceeds) the dot-com/internet boom, while the debt-heavy, opaque financing more closely echoes the subprime crisis or the telecom debt binge with in the broader late-1990s tech cycle.

Financing Mix : Equity vs. Debt

  • Dot-com era (late 1990s–2000) : Heavily equity-financed. Much of the speculation involved internet software/services startups with light physical assets, funded by venture capital and public equity offerings. Losses fell mainly on shareholders (many stocks fell 60–90%). The associated telecom infrastructure build-out did involve substantial debt (~$500+ billion in U.S. telecom bonds/capex in the late 1990s, peaking with sector debt around $300–1 trillion by the early 2000s), including vendor financing (equipment makers like Cisco lending to customers). Still, the overall tech boom’s equity dominance helped limit systemic fallout—the recession was mild.
  • AI today : Shifted rapidly toward debt and off-balance-sheet commitments as hyperscaler free cash flow turned negative. On-balance-sheet debt for major AI firms is heading past $800 billion by end-2026; off-balance-sheet leases, take-or-pay contracts, residual-value guarantees (especially on GPUs), and SPVs exceed $3–4 trillion in notional value. Circular financing is prominent (e.g., Nvidia equity stakes/guarantees enabling customers to buy its chips). Private credit plays a larger role. This creates more leverage, maturity mismatches (short-lived chips vs. longer debt), and opacity—features the ECB links more to subprime than pure equity-funded tech.

Key implication : Dot-com losses were mostly equity wipeouts. AI debt raises the chance of credit losses, forced sales, and wider spillovers if revenues lag.

Scale of Investment Relative to the Economy

AI infrastructure spending is larger as a share of GDP than the internet/telecom boom :

  • Dot-com/internet boom (roughly 1996–2003) : ~1.1% of U.S. GDP annually on average.
  • AI projections (2025–2032) : Often estimated at 3–3.6%+ of U.S. GDP annually (data centers, power, chips), exceeding even the 19th-century railroad boom in some measures (~2.2%). Hyperscaler capex alone is on track for $1.2–1.4 trillion in 2027.

Much AI hardware is imported, so the direct domestic GDP boost is smaller than headline spending, but the financial claims are still large.

Nature of the Assets and Circular Risks

  • Dot-com : Mix of pure software/platforms (low tangible assets) and telecom fiber/networks. Overcapacity left “dark fiber” that was later reused cheaply. Vendor financing created some circularity.
  • AI : Extremely capital-intensive physical build-out (GPUs with short useful lives of ~1–3 years economically, data centers, power). Residual-value guarantees and supplier-backed financing amplify circularity and collateral risk if secondary GPU markets weaken. Concentration is higher (a handful of hyperscalers + Nvidia dominate).

Valuations and Market Concentration

Both eras saw elevated tech valuations and concentration. Dot-com peak S&P 500 CAPE/P/E was higher in some metrics, today’s top AI names form an even larger share of the index (~40% for the top 10 vs. lower in 2000). However, current leaders generate substantial real earnings and cash flows from non-AI businesses, unlike many unprofitable 1999–2000 internet pure-plays.

Potential Consequences of a Correction

  • Dot-com bust : Sharp equity crash, mild recession, limited banking-system damage because equity absorbed most losses. Productivity gains from the internet continued.
  • AI scenario : A pure equity correction could resemble a milder version of 2000–2002. But heavy debt + opacity raises the risk of credit-market stress, project cancellations, and broader tightening—more subprime-like transmission, though hyperscalers’ balance-sheet strength and diversified revenues provide buffers that weaker telecom CLECs lacked. Regulators (Fed, ECB) flag this as a financial-stability watch item rather than an imminent crisis.

Summary Table of Key Differences

Aspect

Dot-Com / Telecom (late 1990s–early 2000s)

AI Boom (mid-2020s)

Primary financing

Mostly equity (VC/public markets) - telecom had heavy debt

Increasingly debt + large off-balance-sheet commitments

Investment scale (GDP)

~1.1% annually

~3–3.6%+ annually (projected)

Asset type

Software + fiber networks

GPUs, data centers, power (short asset lives)

Circularity

Present (vendor financing)

More pronounced (supplier guarantees/equity)

Loss absorption

Mainly equity holders

Equity + credit/private markets

Recession impact

Mild

Potentially broader if credit tightens

Bottom line : The AI debt build-up shares the overinvestment optimism of the dot-com era but carries greater leverage and opacity risks that make pure equity-style outcomes less certain. A correction could still be manageable given the strength of the core players, but the shift to debt financing is the feature that most distinguishes it—and the one regulators are watching most closely. Outcomes will hinge on whether AI revenues materialize at the scale currently priced into both equity and the debt supporting the infrastructure.


Analyze AI Debt Risks.

AI debt risks center on the massive, increasingly leveraged and opaque financing of the AI infrastructure boom (primarily data centers, GPUs, power and related equipment). As of late 2026, this has shifted from largely cash-flow-funded expansion by hyperscalers to heavy reliance on debt and complex off-balance-sheet structures. Regulators (Fed, ECB) and analysts (Goldman Sachs, The Economist, IMF) highlight elevated risks of a correction if AI revenues disappoint with potential spillovers to credit markets, banks and the broader economy—though a full systemic crisis is viewed as less likely than a sharp equity/credit repricing.

Scale of the Debt Build-Up

  • On-balance-sheet debt : The seven largest AI-related firms (Amazon, Alphabet, Meta, Microsoft, Nvidia, Oracle, Broadcom, and similar) are expected to exceed $800 billion in on-balance-sheet debt by end-2026. Hyperscalers alone issued roughly $296 billion in 2026 (above earlier forecasts of ~$250 billion) with Goldman Sachs projecting $420–440 billion in issuance for 2027.
  • Off-balance-sheet and contingent commitments : Far larger—Fed estimates >$3 trillion in maximum notional value (leases not yet started >$1 trillion + take-or-pay contracts and residual-value guarantees >$2 trillion). ECB estimates exceed $4 trillion for a similar group of firms. Goldman Sachs puts key hyperscalers’ lease + purchase commitments around $3.2 trillion.
  • Broader AI-related debt : Advanced-computing debt issuance exceeded $680 billion by October 2026 (vs. $130 billion in all of 2023), accounting for ~31% of nonfinancial investment-grade corporate issuance. Total AI-linked debt exposure is estimated in the high hundreds of billions, with private credit and project finance adding more. Goldman projects significant project finance growth into 2027 (hundreds of billions for shells and chips).

Hyperscalers’ free cash flow turned negative for most in 2026 as AI capex outstripped operating cash flow. Capex for the top players is forecast at $1.2–1.4 trillion in 2027.

Key Risk Factors

1.  Opacity and complexity : Much financing uses special-purpose vehicles (SPVs), long-term leases, residual-value guarantees (especially on GPUs) and take-or-pay contracts. These keep obligations off main balance sheets initially. Risk location is hard to track—held by high-yield funds, insurers, private credit, BDCs, and banks (largest U.S. bank holding companies have ~$1 trillion exposure to tech, data centers, SPEs, private credit and utilities).

2.  Circular and interconnected financing : Nvidia (and to a lesser extent others like Alphabet/Broadcom) provides equity stakes, capacity guarantees, and residual-value backstops (Nvidia’s contingent liabilities estimated in the hundreds of billions, including large deals tied to OpenAI and neoclouds). Customers buy chips/capacity with financing supported by the supplier, creating feedback loops. If demand softens, equity losses, debt impairments and revenue shortfalls can hit simultaneously.

3.  Asset-liability mismatch : AI chips/servers have short economic lives (often 1–3 years estimated by market intelligence) vs. longer debt/lease maturities. Accounting lives have been extended (e.g., to 6 years), but rapid technological obsolescence or overcapacity could impair collateral values.

4.  Revenue realization lag and demand risk : Huge upfront capex requires substantial future AI revenues (e.g., estimates of hundreds of billions annually needed for reasonable returns). Enterprise AI spending has shown signs of plateauing in some data; slower adoption, open-source competition, energy/grid constraints or regulatory pushback (e.g., data-center restrictions) could leave capacity underutilized.

5.  Weaker credits and neoclouds : Pure-play data-center operators and neoclouds (e.g., CoreWeave) carry higher leverage and face rising yields/spreads. Some bond yields have reached distress levels; CDS imply elevated default probabilities. Banks and investors have grown more cautious post-market volatility.

6.  Interest rates, refinancing, and credit conditions : Higher rates squeeze coverage. Spreads on AI-related bonds have widened relative to peers. A broader credit tightening or recession could amplify problems.

Potential Transmission and Severity

  • Direct impacts : Equity correction in AI/tech stocks (already highly concentrated), credit losses for private credit/insurers/banks holding the riskier tranches/project delays/cancellations.
  • Broader effects : Wealth effects on consumers, reduced investment, and possible spillover to utilities/power markets. ECB has drawn parallels to subprime (opacity, off-balance-sheet leverage, circularity) more than pure equity-funded tech booms like the late-1990s dot-com era (which caused a shallower recession partly because equity absorbed more losses).
  • Mitigants : Hyperscalers retain strong overall credit profiles and diversified cash flows from non-AI businesses. Many obligations are contingent or supported by investment-grade parents. Physical assets (buildings, some power infrastructure) retain value even if GPUs depreciate. Contagion to the core banking system appears more limited than in 2008 so far, though private credit opacity is a concern. The Economist and others note a manageable equity crash is more plausible than a deep financial crisis.

Outlook and Monitoring Points

Risks are rising into 2027 as debt issuance is expected to accelerate further and more commitments convert to funded liabilities. Key watch items include :

  • Actual vs. expected AI revenue growth and utilization rates.
  • Further widening of credit spreads or rating actions.
  • Capex guidance and free-cash-flow trajectories from hyperscalers.
  • Regulatory scrutiny (Fed Financial Stability Report already flags this, ECB similarly concerned).
  • GPU secondary-market prices and residual-value realizations.

In summary, AI debt has scaled rapidly into the trillions when including contingent exposures, introducing classic leverage, maturity, and opacity risks on top of technology execution uncertainty. While the strongest players can absorb setbacks, a significant disappointment in AI monetization could trigger a meaningful market correction with credit-market amplification—exactly the dynamic The Economist highlighted in its 2027 outlook. The situation remains fluid and data-dependent.


Why the Economist Predicts a Global Market Crash in 2027.

The Economist does not predict a definite global market crash in 2027. In its annual The World Ahead 2027 outlook (and related leaders), it flags elevated risks of an AI-related equity correction or crash, plus broader pressures from government debt and bond markets that could contribute to financial stress or a downturn. It stops short of forecasting a full-scale global financial crisis.

Main reasons highlighted for AI crash risk

The core concern is financing, not the technology itself advancing too slowly :

  • AI firms (especially those building data centers and infrastructure) are taking on large amounts of debt through increasingly complex and opaque deals. This raises the odds of a sharp correction if returns disappoint or if capital becomes harder to access.
  • Valuations are stretched : AI-exposed stocks form a large share of major indices (e.g., the top AI-related names making up a big portion of the S&P 500). Historical parallels (dot-com era and other technology manias) show investors often overestimate near-term revenues from transformative technologies.
  • External pressures are mounting : political backlash (especially after Democratic midterm gains targeting data centers and certain tech firms), cybersecurity risks from more capable models and potential supply-chain issues.

The Economist explicitly notes that a dot-com-style equity-market crash is plausible, but a deep systemic financial crisis is much less likely. Much of the AI capital spending goes into imported chips/equipment, so the direct hit to the broader U.S. (or global) economy would be smaller than the stock-market impact suggests. Consequences would likely be manageable compared with past crises.

Related pressures that could amplify problems

  • Bond-market and fiscal stress : Rich-country governments are running large deficits. Rising yields (driven by fiscal profligacy, AI-related construction demand for capital, and possible renewed inflation from energy prices) are raising borrowing costs. France is singled out as particularly vulnerable (debt concerns + possible political shift toward Marine Le Pen). The magazine’s view is that “this can’t go on forever—something, somewhere, has to break.”
  • Spillover effects : An AI equity sell-off combined with higher government bond yields could tighten credit more broadly (a risk also noted by the IMF in related coverage). Geopolitical factors (Iran war fallout affecting energy/fertilizer/food prices, Taiwan tensions) add further uncertainty.

In short, the prediction centers on AI financing risks creating the conditions for a significant market correction in 2027, layered on top of already elevated sovereign debt pressures. It is framed as a risk to watch rather than a high-confidence forecast of a global crash. The full World Ahead 2027 package and the November 2026 leader “Why you should expect an AI crash” (subtitled “But not necessarily a financial cataclysm”) are the primary sources.

Thursday, November 19, 2026

Analyze BRICS (Brazil, Russia, India, China and South Africa) Gold settlement plans.

BRICS (Brazil, Russia, India, China and South Africa) has no operational gold settlement system or gold-backed currency. Official plans remain focused on practical local-currency trade settlements and payment interoperability. Gold features mainly in exploratory proposals and national strategies, not as a bloc-wide mechanism.

Official BRICS Position (as of late 2026)

Recent summits (Kazan 2024, Rio 2025, New Delhi September 2026) emphasize :

  • Expanding trade and investment settlements in national currencies.
  • Improving cross-border payment efficiency, interoperability of messaging systems, and infrastructure (BRICS Payment Task Force, discussions of BRICS Pay / BRICS Clear / BRICS Bridge).
  • These are described as voluntary, complementary to existing systems and “no one-size-fits-all.”

Indian officials (hosting the 2026 summit) explicitly stated there is no proposal for a common BRICS currency. Declarations do not mention gold-backed instruments. Progress is incremental and technical rather than revolutionary.

The Main Gold-Related Proposal : “The Unit”

The most discussed concept is a digital settlement unit (often called “The Unit” or UNIT), not a circulating currency people can spend.

  • Design : Roughly 40% physical gold (valued by weight) + 60% basket of BRICS currencies (typically equal weights of the original five: real, yuan, rupee, ruble, rand). Intended as a neutral unit of account or bridge asset for inter-bank / central-bank settlements of trade, reducing the need to convert through the dollar.
  • Status : Originated from Russian research circles (notably IRIAS). Small pilots/prototypes occurred in late 2025 (tiny volumes, e.g., 100 units tested on blockchain). Draft frameworks for a blockchain-based settlement token have been prepared according to some reports, with discussion expected to continue under India’s presidency. It has not been adopted by BRICS governments, central banks collectively, or the New Development Bank as official policy.
  • Purpose : Act as a settlement layer between national systems rather than replace domestic currencies or create monetary union. It aims to handle trade imbalances more smoothly without forcing large holdings of each other’s fiat currencies.

This is frequently overstated in media and social commentary as an imminent “gold-backed BRICS currency.” Fact-checks across multiple summits confirm it remains a prototype/idea under discussion, not an launched system.

Related Gold and Settlement Developments

  • National gold accumulation : BRICS members (especially China and Russia) have been major central-bank gold buyers since 2022. This builds reserves that could support future gold-linked mechanisms and provides sanctions resilience but it is primarily national policy not coordinated BRICS settlement architecture.
  • China’s parallel infrastructure : Stronger and more advanced than bloc-wide efforts. Includes the Shanghai Gold Exchange, yuan-denominated gold contracts, expanding Hong Kong vaulting/clearing (part of a broader “Gold Road” concept), and near-complete local-currency settlement with Russia.
  • Bilateral and technical progress : Local-currency settlement systems (e.g., India-UAE), CBDC pilots (mBridge involving some members and partners) and messaging alternatives. These are more concrete than any gold unit.
  • Occasional talk of a BRICS precious-metals exchange or gold-referenced tools has not advanced to implementation.

Feasibility and Challenges

Advantages of a gold element : Provides a neutral, non-sovereign anchor that is hard to sanction or freeze; reduces pure fiat risk among members with different inflation/policy regimes, aligns with the heavy physical gold stockpiling already underway.

Major obstacles :

  • Divergent interests — India prioritizes pragmatism and avoids confrontation with the dollar system, China prefers advancing the yuan, Russia (under sanctions) pushes harder for alternatives.
  • Technical and liquidity issues — Ensuring enough physical gold backing, custody, valuation (gold price volatility), netting, and trust in a shared ledger.
  • Political and institutional barriers — Creating shared governance without ceding monetary sovereignty is extremely difficult.
  • Scale — Even successful local-currency settlement remains a fraction of global trade, a gold-linked unit would start small.

A recent independent study of BRICS+ currency/payment options concluded that connecting national payment systems and expanding local-currency settlement is far more feasible than any common currency or ambitious shared unit.

Outlook

BRICS is building parallel, multipolar financial infrastructure step by step — payments messaging, local-currency clearing, development banking in local currencies, and national gold reserves — rather than launching a gold standard or single currency. Gold serves as strategic insurance and a potential reference asset, not the core of current settlement plans.

Any gold-linked settlement tool (like an evolved Unit) would likely remain limited to institutional/central-bank use for specific trade corridors if it advances at all. Expect continued bilateral deals, technical pilots and gold accumulation far more than a sudden bloc-wide gold settlement system. Claims of an imminent gold-backed BRICS currency replacing the dollar significantly overstate the evidence.

The Real Reason Behind China’s Non-Stop Gold Buying.

The real reason China keeps buying gold non-stop is strategic insurance against US financial power — and the long-term ambition to reduce the dollar’s dominance.

Official story vs. what’s actually happening

The People’s Bank of China reports steady monthly purchases (now 24 consecutive months through October 2026, with official holdings around 2,435 tonnes). But multiple independent analyses — from Goldman Sachs, Societe Generale and researchers tracking physical flows — indicate China is buying far more than it discloses, potentially 2–10x the official numbers in recent years. Some estimates put true monetary gold holdings closer to 5,000 tonnes. Buying is deliberately opaque to avoid spiking the price further and to limit geopolitical signaling.

This is not casual portfolio tweaking. It accelerated sharply after 2022, when Western countries froze roughly $300 billion of Russia’s central-bank reserves. That episode proved that dollar (and euro) assets held abroad can be switched off overnight. Gold sitting in Chinese vaults cannot.

Core drivers

  • Sanctions and confiscation risk : Gold is the only major reserve asset that is physical, bearer, and not a claim on any government. In a serious US-China confrontation (Taiwan scenario is frequently discussed in this context), China’s massive Treasury holdings and dollar deposits become liabilities. Gold does not.
  • De-dollarization as national security : China has been steadily reducing US Treasury holdings (to multi-year lows). Gold is the preferred replacement because it carries no counterparty risk from Washington. Analysts describe the motive as explicitly political.
  • Building parallel financial infrastructure : China is developing a “Gold Road” / yuan-gold trading network — including offshore vaults (Hong Kong first, then broader Global South/BRICS partners). The goal is to make the yuan more usable internationally by linking it to gold convertibility and settlement outside the dollar system. Higher official gold reserves also bolster confidence in the yuan as a sovereign currency.
  • Domestic reinforcement : China’s property market collapse left households with few good stores of value. Strong private demand for bars and coins (record imports >1,000 tonnes in the first eight months of 2026) complements the central bank’s accumulation and keeps physical gold flowing into the country.

What it is not

It is not preparation for an immediate gold-backed yuan or a formal BRICS currency. Those ideas circulate but the practical focus is resilience and gradual erosion of dollar privilege rather than a sudden monetary revolution. China is accumulating quietly while prices are elevated, treating dips as opportunities.

Bottom line : China is converting paper claims on the US into physical metal it fully controls. The Russia freeze was the wake-up call, sustained US-China rivalry is the ongoing reason. Everything else (reserve diversification, yuan internationalization, household safe-haven demand) is secondary or reinforcing. The buying is deliberate, multi-year and likely to continue.