sample
anish das에 의해budgets, and space programs combined. That single fact captures the magnitude of the shift taking place. Capital is being re-routed from extraction toward cognition. For the first time in human history, the biggest line item on the global balance sheet is not energy, not housing, not transport, but intelligence itself. Late 2025 marks the moment when we crosThis year the world will spend more than four hundred eighty billion dollars building artificial-intelligence infrastructure, more than all global oil exploration, defense sed that threshold, and in hindsight economists will identify it as the starting point of the Intelligence Super-Cycle.
Every major bull market in modern history has been fueled by a collapse in some fundamental cost. The industrial boom followed the collapse in the cost of energy. The post-war consumer boom followed the collapse in the cost of manufacturing. The digital boom followed the collapse in the cost of communication. The next multi-decade expansion is coming from a collapse in the cost of computation and reasoning. In 2018 it cost about ten dollars to process a billion tokens of language through a large model. In 2025 the same task costs less than three cents. Compute cost has fallen more than ninety-nine percent in seven years. When a capability becomes that cheap, it doesn’t just improve margins; it rewrites economics.
Corporations are responding with capital intensity not seen since the 1950s highway build-out. Global data-center capacity is doubling every thirty months. McKinsey estimates the world will spend roughly five trillion dollars on AI-related compute, cooling, and power infrastructure by 2030. U.S. corporate AI capital expenditures are growing at a forty percent compound rate, easily the fastest growth category in the S&P 500. The productivity response is already visible. Non-farm business output per hour in the United States rose 2.7 percent in the past year, the strongest reading in two decades, and roughly half of that gain can be traced to machine-learning deployment inside logistics, finance, and professional services.
Analysts who worry that this is 1999 again miss two crucial distinctions. First, the revenues are real and growing. Second, the balance sheets are strong. In 1999 the median tech company carried net debt equal to thirty percent of assets. Today the median AI-exposed company sits on net cash. The top seven tech firms collectively hold more than seven hundred billion dollars in cash and short-term investments. They are funding this expansion from profits, not leverage. The dot-com era was financed by venture capital; the AI era is financed by free cash flow.
The earnings picture confirms it. Over the past twelve months, aggregate earnings for the information-technology and communication-services sectors rose twenty-three percent year-on-year, while total S&P earnings were up just seven. Profit margins for the large-cap technology cohort have reached a record twenty-six percent. Productivity at that scale acts like a national growth engine. Each percentage-point increase in tech margins adds about eighty basis points to U.S. corporate profits overall. Put differently, AI investment is already responsible for roughly one-third of total earnings growth in 2025.
Globally the story is similar. South Korea’s semiconductor exports are up sixty percent from last year. Taiwan’s fab utilization is above ninety-five percent. Europe’s AI software exports have doubled in two years. India’s information-technology services revenue is growing at twenty-eight percent as Western companies offshore AI implementation work. The International Monetary Fund now estimates that generative-AI adoption could lift global GDP growth by 1.4 percentage points annually through 2035. That may sound small, but applied to a one-hundred-trillion-dollar world economy it equates to roughly thirty-five trillion dollars of incremental output—equivalent to adding another United States plus another Japan.
Valuation concerns remain, but the math looks different once productivity accelerates. The forward price-to-earnings ratio for the S&P 500 is about twenty-one, well below the thirty-plus levels that prevailed during prior euphoric peaks. More important, earnings revisions are positive in two-thirds of industries for the first time since 2018. Historically, when revisions are broad and liquidity conditions stable, markets compound between twelve and fifteen percent annually for several years. The monetary backdrop supports that pattern. Real rates are near neutral, inflation expectations anchored, credit spreads contained. Liquidity is not booming but it is expanding, and expansion at a time of innovation is precisely the environment that breeds sustained bull markets.
Skeptics argue that capital expenditures are outpacing revenues. In nominal terms they are right—AI capex is rising about twice as fast as AI-related revenue. But the same was true for railroads in the 1850s, electricity in the 1890s, and broadband in the 1990s. Infrastructure build-outs always front-load investment because capacity must precede demand. History shows that once utilization passes forty percent, profit curves steepen dramatically. Cloud utilization for AI workloads has already reached thirty-five percent globally and is expected to cross fifty by 2027. That inflection is the moment when cash-flow growth accelerates and equity multiples expand, not contract.
Another overlooked data point is corporate efficiency. U.S. companies adopting AI at scale report an average thirty-percent reduction in administrative cost and a twenty-percent reduction in time-to-market for new products. Those are not hypothetical savings; they are quantifiable improvements in return on invested capital. Across the S&P, ROIC has risen from 10.8 percent in 2022 to 12.6 percent today, the fastest increase in over a decade. That two-point change, if sustained, justifies about a fifteen-percent higher equity valuation even with no further earnings growth.
On the employment side, the data contradicts the fear narrative. Since 2020 the number of high-skill jobs requiring machine-learning familiarity has tripled, and total employment in the U.S. technology sector stands at an all-time high of 9.3 million. The Bureau of Labor Statistics projects AI-related professional occupations will grow twenty-six percent through 2032, roughly four times the national average. Automation displaces some tasks, but it expands the economic pie faster. Historically, every one percent rise in productivity correlates with a 0.5 percent rise in real wages within three years. Early wage data from finance, healthcare, and manufacturing show that pattern repeating.
Energy demand is often cited as a constraint, but it may become an accelerant. AI data centers consume roughly three hundred terawatt-hours of electricity today—less than two percent of global generation—and utilities are already investing to meet projected demand of nine hundred terawatt-hours by 2030. Renewable-capacity additions are running at record levels, and grid-modernization spending is up fifty percent year-on-year. The incremental power required by AI is spurring innovation in storage, nuclear small-modular reactors, and high-efficiency cooling. The net effect could be an energy renaissance rather than an energy crisis.
From a market-structure standpoint, the composition of the rally is healthier than most realize. In 2020 seven companies accounted for more than seventy-five percent of the S&P’s gains. In 2025 that share has fallen to about fifty-five percent as industrials, energy, healthcare, and financials adopt AI and capture their own margin expansion. Breadth is improving. More sectors are participating in the profit up-cycle, and that diversification tends to prolong bull markets. In the past fifty years, whenever earnings breadth widened for three consecutive quarters, the S&P delivered positive returns the following year one-hundred percent of the time, averaging thirteen percent gains.
Look also at global participation. Market capitalization outside the United States now represents forty-nine percent of world equity value, up from forty-two percent five years ago. Emerging-market technology and manufacturing stocks are gaining share as supply chains regionalize. Nations investing early in AI education and cloud infrastructure—India, Indonesia, Vietnam, Poland, Brazil—are reporting productivity gains between 2 and 4 percent annually. That convergence is likely to stabilize growth and reduce volatility in global demand, another tailwind for long-term investors.
Liquidity underpins everything. Global M2 money supply is expanding about 5 percent year-on-year. Corporate bond issuance has normalized, with spreads near pre-pandemic averages. Venture and private-equity funding, which froze in 2022, has rebounded to a twelve-month running total of $780 billion. Roughly sixty percent of that is directed toward AI, automation, and infrastructure plays. Capital availability is the oxygen of innovation, and the world has plenty of it. That is why downturns so far have been brief and self-correcting.
The valuation of intangible assets—software, data, algorithms—is another under-appreciated element. In 1990 intangibles represented about 30 percent of total corporate value. Today they represent nearly 70 percent. Accounting standards haven’t caught up, which means book values understate real enterprise value. When investors eventually mark those assets correctly, price-to-book ratios will normalize and many supposedly “expensive” firms will appear reasonably priced. This is the same adjustment that occurred during previous technological transitions.
Consider the household sector. In the United States, household net worth stands above one-hundred-fifty trillion dollars, up nearly thirty trillion since 2020. Cash holdings exceed six trillion. Mortgage debt as a share of disposable income is the lowest since the late 1990s. The consumer balance sheet can absorb shocks and support demand for goods and services created by the AI economy. That resilience provides the income base for corporate revenues and thus for equities. History shows bull markets end when consumers are over-leveraged, not when they are under-leveraged.
One often-missed dimension is government participation. Public-sector digital-transformation budgets in the G20 now exceed one trillion dollars annually. Defense, healthcare, taxation, and logistics agencies are among the largest AI customers. Because these contracts tend to be multi-year and inflation-indexed, they provide stability that private markets lack. In effect, governments are underwriting a portion of the AI build-out. That base demand smooths cycles and gives corporations confidence to invest aggressively.
Now, regarding valuation fears about specific companies: yes, some trade at high multiples, but the outliers are driving system-wide efficiency. If a firm is priced at one hundred times earnings yet enables dozens of clients to raise margins by five points, its social return on capital justifies the headline ratio. Market mechanisms eventually distribute those gains through suppliers, employees, and investors. That’s why periods of perceived overvaluation often precede long expansions rather than end them.
Long-term equity performance still hinges on three inputs: earnings growth, dividend yield, and multiple expansion. The dividend contribution in this cycle may be modest, around 1.5 percent annually. But with real GDP growth near 3 percent and productivity adding another 1.5, nominal earnings could compound at 8 to 10 percent. Add even slight multiple expansion as inflation stabilizes, and total returns in the low-teens become plausible. That’s not euphoria—that’s arithmetic.
Another way to view the cycle is through capital formation. In the last five years, global equity issuance averaged $1.2 trillion annually, well below the 2010s average of $1.8 trillion. Companies have been buying back more shares than they issue, reducing float by about one percent per year. Meanwhile, profit pools are growing. Fewer shares and larger earnings mean higher per-share value, even without valuation expansion. Buybacks are expected to exceed one trillion dollars in the U.S. again this year. That mechanical tailwind alone adds roughly three percentage points to total equity returns.
From a behavioral standpoint, sentiment is still cautious. Surveys show retail cash allocations near post-crisis highs. The American Association of Individual Investors bullish reading is just 37 percent, well below levels that typically precede bubbles. Institutional positioning is defensive, with average equity exposure among global balanced funds below historical norms. Bull markets rarely end when investors are skeptical; they end when everyone is all-in. We are not there yet.
The demographics favor expansion too. Over six hundred million millennials and Gen Z workers worldwide are entering their peak earning years. They are digital natives comfortable using AI tools. Their productivity curve is steeper, and their consumption patterns favor digital goods and experiences with high profit margins. This cohort will drive both supply and demand in the AI economy, reinforcing the cycle.
Inflation dynamics are improving as well. AI-driven supply-chain optimization is cutting logistics costs by fifteen percent, and predictive maintenance in manufacturing reduces downtime by twenty. Those efficiencies feed directly into lower unit costs, offsetting wage pressure. Disinflation anchored by productivity, not austerity, is the healthiest kind of macro environment. It allows central banks to maintain moderate rates, supporting investment valuations without overheating.
All these strands—productivity, liquidity, demographics, sentiment, valuation—interlock into one conclusion: the conditions for a sustained equity expansion are in place. This is not 1999; it is closer to 1954, the beginning of a long era when technological diffusion and capital abundance coexisted. The AI boom may experience corrections, but the secular trend points higher because the underlying economics point higher.
The transition we are witnessing will not end with chatbots or image models. It will permeate energy management, healthcare diagnostics, materials science, and finance. Every sector that digitizes decision-making gains scale without proportional cost, and that asymmetry is what drives wealth creation. Markets price that asymmetry long before accountants record it. That is why staying structurally bullish on innovation has historically paid.
The final metric worth remembering is time. Since 1950 there have been eighteen distinct bear markets in the S&P 500. The average decline lasted 14 months; the average expansion lasted 64. On that ratio, investors spend roughly four out of every five years compounding. The lesson is simple: secular trends dominate cyclical noise. Artificial intelligence is a secular trend measured in decades, not quarters.
As we close 2025, global equity market capitalization stands near one-hundred-fifteen trillion dollars, up twelve trillion in two years. The combined revenue of the world’s thousand largest public companies exceeds sixty trillion and is growing at nearly ten percent annually. Those numbers are the scoreboard of capitalism adapting to a new factor of production: synthetic intelligence. The economy is not running out of ideas; it is learning how to commercialize ideas faster.
Looking forward, the next decade’s winners will be defined by data leverage, operational efficiency, and capital discipline. Firms that align those three will compound. The rest will consolidate. For investors and policymakers alike, the priority is to understand that this expansion is not speculative excess but structural modernization. The world is reallocating capital from consumption to cognition, and that process, historically, has created prosperity.
The conclusion is straightforward. The AI era will be remembered not only for transforming industries but for revitalizing productivity and profitability at a scale capable of lifting global living standards. Equity markets reflect that promise. They may fluctuate, but the direction of progress remains up and to the right. In every generation there comes a moment when fear of change yields to acceptance of growth. That moment is now.
