OpenAI, Oracle, and Nvidia have constructed an unprecedented financial loop worth over $400 billion where capital flows in circles—Nvidia invests $100 billion in OpenAI, which commits $300 billion to Oracle for cloud services, which then spends $40 billion buying Nvidia chips.
Fortune +6
This arrangement, which analysts have dubbed an “infinite money glitch,” allows all three companies to simultaneously inflate their valuations while the same dollars circulate through the system.
Substack +4
The pattern mirrors the circular vendor financing that characterized the late 1990s dot-com bubble, where telecom equipment makers like Cisco and Nortel funded their own customers before everything collapsed in 2001.
Yahoo Finance +2
Yet unlike that era, today’s AI leaders generate substantial real revenue—OpenAI hit $12 billion in annual recurring revenue by mid-2025,
CNBC
while Nvidia posted $130.5 billion in total revenue with extraordinary 53% net margins.
Next Platform +2
The critical question is whether these arrangements represent genuine demand for transformative technology or an elaborate shell game that will unravel when the music stops.
As of October 2025, 54% of fund managers believe AI stocks are in bubble territory—the first time “AI equity bubble” has been identified as the top global tail risk in Bank of America’s survey history.
EY
Yahoo Finance
The ecosystem now captures nearly 50% of all global venture capital,
R&D World
has spawned 498 unicorns worth $2.7 trillion combined, and is driving $4-8 trillion in projected infrastructure spending through 2030. Whether this represents rational investment in the future or irrational exuberance depends entirely on whether AI companies can generate enough revenue to justify their astronomical valuations and meet their massive financial obligations.
The architecture of circular finance
At the heart of the AI boom sits a complex web of cross-investments, purchase agreements, and equity stakes that create what critics describe as artificial demand. Nvidia announced plans to invest up to $100 billion in OpenAI in September 2025, structured as progressive investments tied to infrastructure deployment.
OpenAI +3
The first tranche of $10 billion would fund the first gigawatt deployment expected in the second half of 2026, with the full investment potentially giving Nvidia approximately 25% ownership of OpenAI at its $500 billion valuation.
CNBC
CNBC
This investment flows directly back to Nvidia through chip purchases—OpenAI is leasing rather than buying AI processors, with most of the $100 billion ultimately returning to Nvidia as revenue over the useful life of the GPUs.
CNBC +2
The pattern extends through Oracle’s massive involvement. OpenAI committed to a $300 billion contract over five years with Oracle starting in 2027, providing 4.5 gigawatts of computing power through the Stargate project.
TechCrunch +5
This deal was later expanded to nearly 7 gigawatts and over $400 billion in investment over three years. Oracle’s CFO Clay Magouyrk defended the arrangement by noting OpenAI’s rapid growth to “almost a billion users,” but critics point out that Oracle must spend heavily to fulfill these contracts.
CNBC
Oracle placed a $40 billion order for Nvidia GB200 GPUs—approximately 400,000 chips—under a 15-year lease agreement to build the Stargate data center in Abilene, Texas.
Trending Topics +4
The 1.2 gigawatt facility across 875 acres will become operational by mid-2026, with Oracle purchasing the chips and leasing computing power back to OpenAI.
TechSpot
The circular dynamics become even more pronounced when examining the broader ecosystem. Nvidia holds a 7% stake in CoreWeave worth approximately $3 billion as of the second quarter of 2025.
Yahoo Finance +2
CoreWeave has purchased at least $7.5 billion in Nvidia hardware—primarily H100 Hopper GPUs at roughly $30,000 each—meaning all of Nvidia’s investment has already returned as revenue.
Yahoo Finance +2
Nvidia then agreed to purchase $6.3 billion in unsold cloud capacity from CoreWeave through April 2032, effectively guaranteeing CoreWeave’s business model while securing compute for its own needs.
CNBC +4
Meanwhile, OpenAI holds a $350 million equity stake in CoreWeave
CNBC
and has committed $22.4 billion over five years for AI data center capacity from the company.
CNBC +3
Dylan Patel of Semianalysis coined the term “infinite money glitch” to describe the mechanism: Nvidia invests capital in OpenAI, which uses those funds to purchase Nvidia hardware directly and buy Oracle cloud compute. Oracle then uses that revenue to purchase Nvidia hardware for its data centers. Nvidia books all of this as revenue despite the revenue being spurred by its original investment. The new revenue supports Nvidia’s valuation, making its stock more valuable, which allows Nvidia to invest even more in the next round.
Substack
Jensen Huang, Nvidia’s CEO, stated that a 1 gigawatt data center costs approximately $50 billion total, with $35 billion going to Nvidia GPUs alone.
CNBC
KrASIA
For the 10 gigawatt project OpenAI is pursuing, that potentially translates to $350 billion in Nvidia hardware over multiple years.
KrASIA
Stacy Rasgon of Bernstein Research warned that this arrangement “will likely fuel circular worries much hotter than what we have seen previously,”
AIM Media House
while Jay Goldberg of Seaport Global Securities compared it to “parents co-signing your first mortgage” and noted the deals carry a “whiff of circular financing” that are “emblematic of bubble-like behavior.”
Yahoo Finance +2
The concern is that when everyone is simultaneously buyer, seller, and investor, massive correlation risk is created. If OpenAI cannot pay Oracle, Oracle cannot pay Nvidia, Nvidia’s stock crashes, and because these companies represent roughly 25-30% of the S&P 500’s total market value, the entire market could enter freefall.
Substack
NBC News
Parallels to the dot-com era, with crucial differences
The current AI boom exhibits striking similarities to the 1995-2001 internet bubble, yet also differs in fundamental ways that complicate any simple comparison. During the dot-com era, venture capital deployment exploded from under $5 billion annually in the early 1990s to over $100 billion in 2000, with 39% of all VC investments flowing to internet companies by 1999.
Britannica +2
Today, AI investment has reached even higher concentration—AI captured $192.7 billion or approximately 50% of ALL global venture capital through October 2025.
Bloomberg
This level of dominance exceeds even the dot-com peak, with AI investment as a share of the economy being one-third greater than internet-related investments during the bubble, according to Sequoia Capital analysis.
goldmansachs
The valuation methodologies show both parallels and divergences. During the dot-com era, companies were valued based on “eyeballs,” page views, and traffic growth regardless of monetization plans. The NASDAQ’s forward price-to-earnings ratio hit approximately 60x at the March 2000 peak, with Cisco reaching a P/E ratio of 196.
IntuitionLabs
Wikipedia
Today’s AI companies use metrics like annual recurring revenue, token usage, and API calls—more grounded in actual business activity. The average P/E ratio for the top seven AI-related companies is approximately 25x, less than half the dot-com peak. Nvidia’s P/E of 52-53x is elevated but nowhere near Cisco’s 150x during the bubble.
goldmansachs
However, revenue multiples tell a more concerning story. AI startups command an average 44.1x revenue multiple for LLM vendors and 30.9x for AI search engines, with some companies like Character AI trading at 568x revenue.
Finrofca
While these are high, they pale compared to Cisco’s 200x price-to-sales ratio at its peak.
Research Affiliates
Goldman Sachs’ Peter Oppenheimer notes that “leading tech companies are less extreme in valuation today, and the broader optimism that spilled over into equity valuations in the late 1990s is not apparent today.”
goldmansachs
Goldman Sachs
The critical difference lies in profitability and revenue generation. Only 14% of dot-com IPOs were profitable in 2000, and most companies had no viable path to profitability or even substantial revenue.
Wall Street Mojo
Finbold
Pets.com burned $300 million in 268 days before liquidating nine months after its IPO, spending 90% of its budget on marketing.
Finbold
Today’s AI landscape is fundamentally different at the infrastructure level—Nvidia generated $130.5 billion in revenue in fiscal 2025 with 53% net margins,
Semiconductor Engineering +2
Microsoft’s AI business exceeded $13 billion in revenue,
Epoch AI
and Google confirmed that demand for AI services nearly tripled between May and October 2024. These are highly profitable businesses with genuine customer demand.
Yet at the AI-native startup level, the parallels are more troubling. OpenAI is projected to lose $14 billion in 2025 despite $13 billion in revenue, with profitability not expected until 2029.
GeekWire
The company’s cumulative losses from 2023 through 2028 are projected at $44 billion. An MIT study found that 95% of 52 organizations achieved zero ROI despite spending $30-40 billion on generative AI across more than 300 initiatives.
yale
This echoes the dot-com era’s fundamental problem: massive investment with minimal return.
The infrastructure overinvestment patterns are remarkably similar. During the dot-com boom, telecoms laid over 80 million miles of fiber optic cable based on WorldCom’s fraudulent claim that traffic was doubling every 100 days. The result was that 85-95% of fiber remained “dark” for years after the crash, with Corning stock plummeting from $100 to $1 and networking equipment companies losing 90%+ of their value.
Fortune
Yahoo Finance
Today, data center capital expenditures reached $134 billion in the first quarter of 2025 alone, up 53% year-over-year,
CIO Dive
with plans for $320 billion in additional spending in 2025.
Fortune
Yahoo Finance
Meta is planning data center infrastructure “the size of Manhattan,” and the Stargate Project commits $500 billion to AI data center networks nationwide.
Fortune +4
If AI adoption proceeds slower than projected, this could create massive overcapacity similar to the fiber optic buildout.
Interest rate environments provide an interesting contrast. The dot-com bubble burst as the Federal Reserve raised rates multiple times in 1999-2000, reaching 6.5% by May 2000.
Fortune
Rising rates made speculative investments less attractive and exposed fundamentally weak business models. Today’s environment features elevated rates but with a trajectory toward gradual cuts, providing a tailwind rather than headwind. This key difference suggests less immediate pressure on valuations from monetary policy.
The circular vendor financing patterns, however, are nearly identical. Before the 2001 crash, the top five telecom equipment makers’ customer financing exceeded 123% of their combined earnings, with Cisco and Nortel’s customer financing exceeding 10% of annual revenues.
Yahoo Finance
Fortune
Global Crossing engaged in direct “revenue roundtripping”—paying companies for services while those companies bought equipment of exactly equal value. When the bubble burst, Global Crossing went bankrupt and networking equipment businesses collapsed.
Yahoo Finance
Fortune
Today’s arrangements between Nvidia, OpenAI, Oracle, CoreWeave, and others follow remarkably similar patterns, raising concerns about whether we’re witnessing history repeating itself.
The great valuation debate and bubble evidence
The question of whether we are in an AI bubble has divided experts, financial institutions, and market participants into sharply opposing camps, each with compelling evidence. The bearish case begins with valuation disconnects that defy traditional financial logic. OpenAI’s valuation doubled from $300 billion to $500 billion in less than a year despite losing $5 billion annually and projecting cumulative losses of $44 billion through 2028.
EY
The company has made approximately $1 trillion in AI infrastructure deals while generating only $13 billion in revenue—
CNBC
a ratio that seems unsustainable without dramatic revenue acceleration.
CNBC
CNBC
Even more extreme is Thinking Machines, which raised $2 billion at a $10 billion valuation in the largest seed round in history without releasing a product or disclosing what they’re building.
Market concentration has reached levels that create systemic risk. Just seven stocks—the “Magnificent Seven”—accounted for 55% of S&P 500 gains since the end of 2022, with AI-related stocks driving 75% of index returns, 80% of earnings growth, and 90% of capital spending growth since ChatGPT launched, according to JPMorgan’s Michael Cembalest.
yale
CNN
When the top 10 S&P 500 companies command such dominance and their valuations depend on AI transformation proceeding on schedule, correlation risk becomes extreme. Yale’s Jeffrey Sonnenfeld warns that “should the bold promises of AI fall short, the dependence among these major AI players could trigger a devastating chain reaction, causing a widespread collapse similar to the 2008 Great Financial Crisis.”
yale
Yale Insights
Goldman Sachs’ Jim Covello, head of global equity research, poses a fundamental question: “What trillion-dollar problem will AI solve? Replacing low-wage jobs with tremendously costly technology is basically the polar opposite of prior technology transitions.” His analysis highlights the vast gap between investment levels and actual credible expectations for future profits, which “certainly looks bubbly.”
Institutional Investor
Bain & Company projects that AI companies will need $2 trillion in annual revenue by 2030 to justify infrastructure spending, yet current trajectories suggest an $800 billion shortfall.
CNBC
The circular financing dynamics amplify concerns about artificial demand. Fortune Magazine notes that “the extent to which the entire AI boom is backstopped by Nvidia’s cash isn’t easy to answer precisely, which is also one of the unsettling things about it.” NewStreet Research estimates that for every $10 billion Nvidia invests in OpenAI, it sees $35 billion in GPU purchases—representing approximately 27% of Nvidia’s annual revenue.
Yahoo Finance
Fortune
Harvard Business Review warns that “speculative capital and circular financing can distort timing and expectations,” while consumer enthusiasm outpaces enterprise integration, raising the risk of overcapacity.
Harvard Business Review
The bullish case, however, rests on fundamentally different premises backed by hard financial data. Nvidia’s earnings multiplied 15-fold from 2020 to 2024, jumping from $0.04 to $0.60 per share in just the first quarter. Revenue grew 62% in its latest fiscal year, reaching $130.5 billion with extraordinary 53% net margins—
Sacra
Next Platform
these are not speculative numbers but actual profitable business at massive scale.
Semiconductor Engineering +2
OpenAI’s annualized revenue leaped from $1 billion to $5 billion in under six months,
SaaStr
while Anthropic achieved similar 5x revenue growth in the same period.
Epoch AI +2
Google confirmed that demand for AI services nearly tripled between May and October 2024, demonstrating accelerating rather than plateauing adoption.
Noahpinion
Goldman Sachs economist Joseph Briggs argues that AI applications are leading to real productivity gains, with expectations that U.S. companies will generate $8 trillion in new revenue from AI.
NBC News
The cost to run a GPT-3.5-level model dropped 280-fold between November 2022 and October 2024, while the price per FLOP/s on Nvidia GPUs fell over 75% in the same period.
NBC News
These dramatic efficiency gains enable new use cases and business models that were previously uneconomical, expanding the addressable market significantly.
Goldman Sachs’ Peter Oppenheimer emphasizes that “we believe we are still in the relatively early stages of a new technology cycle that is likely to lead to further outperformance.” He notes that tech leaders’ cash as a percentage of market cap is double what companies had during the internet bubble, while return on equity and average margins are nearly double 1990s levels. Tech sector profit margins currently stand at 26%—more than double 2004 levels—meaning these companies are generating real profits, not just inflated expectations.
goldmansachs
Goldman Sachs
JPMorgan CEO Jamie Dimon offers a nuanced perspective: “You can’t look at AI as a bubble, though some of these things may be in the bubble. In total, it’ll probably pay off.” He compares AI to the internet, which led to a dot-com bubble but ultimately created real economic transformation.
NBC News
NBC News
This view acknowledges that while specific companies and valuations may be excessive, the underlying technology has genuine value that will materialize over time.
The adoption data supports the optimistic view. Business AI adoption jumped from 55% in 2023 to 78% in 2024, while generative AI adoption surged from 33% to 71% in the same period according to Stanford University research.
Stack AI
One in three companies globally is planning to invest $25 million or more in AI in 2025.
Moonfare
Global AI industry revenue is expected to grow from $279 billion in 2024 to $3.5 trillion by 2033—a 32% compound annual growth rate.
Stocktitan
These figures suggest widespread conviction that AI will transform business operations, not just hype-driven experimentation.
The moderate view, held by many sophisticated investors, acknowledges bubble characteristics while expecting correction rather than catastrophe. David Solomon, Goldman Sachs CEO, predicts a “drawdown” in 12-24 months but not a complete collapse.
CNBC
Jeff Bezos called the current environment “kind of an industrial bubble” where “every experiment gets funded” but expects sorting between winners and losers rather than total failure.
Yahoo Finance
Sam Altman himself acknowledged that “people will overinvest and lose money” during this phase,
TechCrunch
while also asserting that AI is “the most important thing in a very long time.”
Fortune +2
Pat Gelsinger, former Intel CEO, offered perhaps the most pragmatic assessment: “Are we in an AI bubble? Of course! We’re hyped, we’re accelerating, we’re putting enormous leverage into the system”—but he expects it will last “several years” rather than imploding immediately.
CNBC
This view suggests that bubble dynamics can persist when underlying technology continues delivering results, even if specific valuations are excessive.
Market realities and the fragility beneath
The current state of AI valuations and market dynamics reveals a landscape of extraordinary opportunity shadowed by systemic vulnerabilities. AI venture capital investment reached $192.7 billion through October 2025, representing approximately half of all global venture capital and a concentration level never before seen in a single technology sector.
Bloomberg
The 498 AI unicorns worth $2.7 trillion combined have created more paper wealth than any previous technology wave, with 100 new unicorns minted since 2023 alone.
Fortune
Yet beneath these staggering numbers lies a more complex reality about profitability, sustainability, and the path forward.
The revenue-to-valuation gap at the startup level remains enormous. Anthropic achieved a $183 billion valuation in September 2025 despite $5 billion in annual recurring revenue—a 36.6x multiple.
Anthropic
CNBC
While impressive, this pales compared to Databricks at $62 billion on roughly $1 billion in revenue,
Industry Leaders Magazine
representing a 62x multiple.
Finerva
CNBC
Cohere raised capital at a $5.5 billion valuation on just $22 million in revenue, implying a 250x multiple. These figures suggest that investors are betting not on current business performance but on winner-take-all dynamics where one or two companies will capture the majority of value creation in each AI category.
The profitability challenge cannot be overstated. OpenAI is burning $8 billion in cash in 2025 despite $12 billion in revenue, primarily due to compute costs. The company lost $5 billion in 2024 and doesn’t expect to reach cash flow positive until 2029, by which time it projects needing $100 billion in annual revenue.
TapTwice Digital
Most AI-native companies face similar dynamics—explosive revenue growth coupled with even more explosive cost growth. An MIT study found that 80% of companies report limited EBIT impact from generative AI, with only 17% seeing 5% or greater profit improvement.
McKinsey & Company
yale
This suggests that while AI creates value for infrastructure providers like Nvidia and cloud platforms, the economic returns for end users remain elusive.
The infrastructure layer tells a different story. Data center capital expenditures reached $134 billion in Q1 2025 alone, up 53% year-over-year, with AWS, Microsoft, Google Cloud, and Meta planning over $250 billion in buildouts for 2025.
CIO Dive
Global data center capacity stands at 62 gigawatts currently, with projections for a 50% surge by 2027. AI workloads are expected to represent 28% of this capacity by 2027, up from just 13% in early 2023. Whether this massive buildout represents foresight or overcapacity depends entirely on whether enterprise AI applications deliver sufficient ROI to justify continued investment.
The M&A market has exploded as companies race to acquire AI capabilities rather than build them organically. M&A activity surged 155% year-over-year with over $100 billion in startup acquisitions in the first half of 2025.
PYMNTS
Crunchbase News
Google is reportedly planning a $32 billion acquisition of Wiz, which would be the largest startup acquisition on record. Meta acquired 49% of Scale AI for $14.8 billion, while OpenAI acquired Jony Ive’s io for $6.5 billion.
PYMNTS
This acquisition spree suggests that major tech companies believe they’re in a race where speed matters more than price, and missing the AI transition poses existential risk.
The IPO pipeline remains selective but active, with companies like Databricks targeting $62-100 billion valuations for listings expected in 2025-2026.
CNBC
Quartr
CoreWeave, which jumped from $2 billion to $23 billion in valuation over 18 months, is targeting an IPO in Q2 2025 with expectations for $35+ billion valuation. These IPOs will provide a critical test of public market appetite for AI companies and whether private valuations can be sustained under public market scrutiny.
Geographic concentration amplifies systemic risk. North America captured 70% of global AI funding in H1 2025, with 29 of 39 recognized AI unicorns based in the United States.
EY +2
Asia hit multi-year lows, with China down 33% year-over-year, though Chinese companies like DeepSeek are demonstrating dramatically lower costs—training costs 18x lower and inference costs 36x lower than Western counterparts.
McKinsey & Company
If this cost efficiency can be replicated broadly, it could undermine the economic rationale for massive Western infrastructure investments.
Enterprise adoption metrics show widespread experimentation but limited transformation. While 70-78% of enterprises use AI in at least one business function,
Stack AI
daily enterprise use remains concentrated in specific sectors. The information sector shows 25% adoption rates while accommodation and food services languish at 2.5%.
Anthropic
Only 5% of ChatGPT’s 800 million users pay for subscriptions, compared to 3% across all generative AI services—
The Register
a conversion rate that raises questions about long-term revenue sustainability from consumer markets.
The energy and power constraints represent a potentially binding constraint on growth. AI data centers require enormous amounts of electricity, with 10 gigawatts enough to power a major city. Meta is planning $600 billion in infrastructure spending through the end of 2028,
TechCrunch
much of it for energy-intensive AI operations. Power generation capacity cannot expand as quickly as AI ambitions, creating bottlenecks that may slow deployment regardless of funding availability. Partnerships between tech companies and utilities are accelerating, but building new power generation takes years, not months.
Navigating between transformation and collapse
The AI financial ecosystem exists in a state of tension between two competing realities that will determine whether current valuations represent prescient investment or catastrophic misallocation. On one side stands genuine technological breakthrough—large language models demonstrating capabilities that seemed impossible just three years ago, with adoption accelerating at rates that dwarf previous technology waves. ChatGPT reached 100 million users in two months, faster than any consumer application in history. Google processed 10 trillion tokens in April 2024, then 980 trillion by June 2024, a doubling every two months that suggests exponential growth in real usage rather than speculative hype.
Semiconductor Engineering
On the other side looms the circular financing apparatus that amplifies both gains and losses through a tightly interconnected web of companies that are simultaneously investors, suppliers, and customers. The mathematics of this arrangement create fragility—if AI adoption slows or profitability remains elusive, $560 billion has been invested by big tech in AI infrastructure over two years while generating only $35 billion in AI revenue combined, a 16:1 investment-to-revenue ratio that looks unsustainable if the gap doesn’t narrow substantially.
The likely outcome is neither the bulls’ scenario of uninterrupted exponential growth nor the bears’ prediction of complete collapse. Instead, the most probable path involves selective correction where weaker business models fail while core technology and leading companies endure. Unlike the dot-com crash where 86% of companies disappeared,
Wikipedia
AI has demonstrated too much real utility across too many applications to vanish. The technology will survive; many of the valuations will not.
Three critical factors will determine how this resolves. First, whether AI companies can solve the profitability puzzle by reducing compute costs faster than they reduce prices. The 280-fold drop in model running costs and 100x decrease in inference costs from 2022 to 2024 provides hope,
Semiconductor Engineering
but this must continue while revenue growth accelerates. Second, whether enterprise AI applications deliver measurable return on investment that justifies continued spending. If the 80% of companies seeing limited EBIT impact becomes 60% or 40% over the next two years, investment will continue.
McKinsey & Company
If it rises to 90%, funding will evaporate. Third, whether the circular financing arrangements unwind gracefully or catastrophically. If OpenAI can grow into its valuation before needing to refinance its massive commitments, the system stabilizes. If it cannot, the correlation effects could trigger cascading failures.
The historical parallel most apt may not be the dot-com bubble but rather the railroad boom of the 1840s in Britain and the 1860s-1890s in America. Railroad companies raised enormous capital, built vast infrastructure on speculation, engaged in financial engineering that enriched insiders while bankrupting many investors, yet ultimately created the foundation for economic transformation. Many railroad companies failed, their investors were wiped out, but the infrastructure they built enabled decades of subsequent growth. As Jim Cramer noted, “once the losers got wiped out, the winners won big.”
NBC News
The AI bubble question thus has multiple correct answers depending on perspective and timeframe. For many late-stage AI startup investors entering at peak valuations, this is almost certainly a bubble that will destroy capital. For shareholders of Nvidia, Microsoft, Google, and other infrastructure providers with profitable AI businesses today, this represents a genuine long-term growth opportunity despite near-term volatility. For the economy as a whole, the massive infrastructure buildout may prove valuable even if the companies financing it fail, just as fiber optic networks eventually found uses after sitting dark for years.
The risk that should concern policymakers and investors most is not whether AI is transformative—it clearly is—but whether the correlation and concentration created by circular financing will amplify a correction into a crisis. When Nvidia represents 7% of the S&P 500, Microsoft another 7%, and the top tech companies collectively account for 25-30% of the index, all interconnected through investment and purchase agreements, a shock to any one company transmits instantly to all the others.
Substack
This systemic vulnerability mirrors the financial crisis of 2008, where interconnections meant that failures cascaded rather than remained isolated.
The next 12-24 months will be critical. If AI companies demonstrate clear paths to profitability, if enterprise adoption deepens rather than plateaus, if infrastructure utilization rises to meet installed capacity, the current valuations may prove justified in retrospect. But if profitability remains elusive, if the MIT study’s finding of 95% zero-ROI projects persists, if demand growth slows while supply continues expanding at exponential rates, the unwinding could be severe and swift. The “infinite money glitch” only works as long as all participants believe it will continue working—once confidence breaks, the circularity that amplified gains will amplify losses with equal force.
Substack
What seems most certain is that we are witnessing a genuine technological inflection point financed through mechanisms that contain both promise and peril. The companies and investors who survive will be those who can distinguish between AI as a technology—which will transform the economy—and AI as an investment theme—which has already priced in decades of future growth that may or may not materialize on the expected timeline. As Rob Arnott observed about the dot-com bubble, “the narrative was correct, but the market bet it would play out faster than it did.”
TraderLion
IntuitionLabs
The same wisdom likely applies to AI: the transformation is real, but the timing and path to value will determine whether today’s investors look prescient or foolish when the story is finally written.