The AI WealthCreation Event

Why this cycle may be larger than the internet — and why the disciplined investor cannot afford to miss it.

Hyperscaler capital expenditure is on track to approach $725 billion in 2026 — nearly double the prior year.¹ NVIDIA has crossed $5 trillion in market capitalization.² Generative AI reached 100 million users in two months — the fastest adoption of any consumer technology in history.³ Every prior wealth-creation event of comparable scale has rewarded investors who participated early with discipline, and punished those who waited for confirmation.

Don Schreiber, Jr. | Principal, WBI Investments | May 2026

Executive Summary: A Generational Opportunity, A Generational Decision

Every two or three decades, the global economy reorganizes itself around a general-purpose technology. Electrification did it. The personal computer did it. The internet did it. Each cycle created enormous, durable wealth — and each cycle was misjudged in real time, dismissed by skeptics until it was too late to participate on favorable terms.

Artificial intelligence is the next of these events, and on most quantitative measures it is unfolding faster, more capital-intensively, and with stronger underlying corporate fundamentals than the internet build-out of 1995–2000. The four largest U.S. hyperscalers have committed to roughly $725 billion of capital expenditure in 2026 alone⁴ — approximately three times their combined 2024 spend, and concentrated overwhelmingly on AI compute, networking, and data centers. NVIDIA, the central supplier to that build-out, has crossed $5 trillion in market capitalization.⁵ ChatGPT reached 100 million users in roughly two months, faster than any consumer product ever measured.⁶

The Three Claims of This Paper

One. The AI cycle exhibits every structural marker of a multi-decade wealth-creation event, and at a scale of capital concentration the United States has not seen in any peacetime cycle in its history. Hyperscaler capex alone reached approximately 2.6% of U.S. GDP in 2026 — twice the peak share of the 1996–2001 telecom build-out — and the broader policy-aligned capital stack pushes the figure higher still.

Two. Unlike the dot-com cycle, the AI cycle is being led by companies that are already among the most profitable enterprises in history. The capital being deployed is not being raised from speculative IPOs — it is being funded out of existing free cash flow at firms with double-digit operating margins.

Three. Industry product capacity to participate intelligently in this cycle remains thin. AI-themed ETFs total a small fraction of the broader thematic-fund universe, and most are passive, market-cap-weighted, and lack any explicit risk discipline. WBI Power Factor AI™ was built specifically to address this gap.⁷

"The cost of waiting until a wealth-creation event is obvious is the same in every cycle. The trade is no longer available on favorable terms."

SECTION 1

The Pattern of Wealth-Creation Events

Wealth-creation events of historic scale share a recognizable signature. They are not driven by a single product, but by a general-purpose technology that reshapes the cost structure of broad swaths of the economy. They exhibit a multi-year capital formation phase, during which infrastructure is built ahead of demand. They reward the suppliers of that infrastructure first, the platform operators second, and the applied users last — and across all three layers, they create durable franchises that compound for decades.

Four Events, One Template

Cycle	Era	Defining Build-Out	Representative Wealth Created
Electrification	1880s–1920s	Power generation, transmission, the modern grid	GE, Westinghouse — multi-decade industrial titans
PC Revolution	1980–1995	Microprocessors, operating systems, productivity software	Intel: ~25,000% return; Microsoft: ~60,000% peak return
The Internet	1995–2010	Fiber, browsers, search, e-commerce, mobile platforms	Cisco peaked >$500B; Amazon, Google compounded for 25 years
Mobile / Cloud	2007–2020	Smartphones, app economy, hyperscale cloud	Apple +$3T; AWS, Azure, GCP collectively the largest IT category
Artificial Intelligence	2022–present	GPUs, foundation models, AI-native data centers	NVIDIA $5T; hyperscaler capex $725B in 2026 alone

Source: Company reports, historical market data

Each of these cycles was identifiable in real time as transformative; each was also accompanied by chorus voices arguing that valuations were stretched, that the technology was overhyped, that prudent investors should wait. In every case, those who waited for unambiguous confirmation participated late, at higher prices, in a smaller share of the durable winners. The skeptic's caution was a real and measurable opportunity cost — paid in compounded returns foregone.

Skepticism is not the wrong instinct in markets. Indiscriminate skepticism is. The investor's task is not to dismiss the cycle, nor to chase it blindly, but to identify the disciplined path of participation.

SECTION 2

The AI Investment Cycle, By The Numbers

The capital formation phase of the AI cycle is the largest, fastest infrastructure build-out in the history of private enterprise. The four largest U.S. hyperscalers — Microsoft, Alphabet, Amazon, and Meta — have collectively guided to capital expenditure of approximately $725 billion in 2026, an increase of roughly 77% from the prior year and nearly triple their combined 2024 spend. Adding Oracle's guidance, the figure exceeds $775 billion.⁸ Independent estimates suggest approximately three-quarters of this is directed at AI-specific infrastructure: GPUs, data centers, networking, and power.

Three Observations That Matter

The capital is real, and it is being deployed. Microsoft alone spent $30.9 billion of capex in a single fiscal quarter — up 84% year-over-year — and confirmed an annual AI revenue run-rate above $37 billion. Alphabet's quarterly capex more than doubled. Amazon committed $200 billion for the full year. These are not projections; they are corporate guidance backed by signed leases, ordered hardware, and disclosed remaining performance obligations.⁹

The capital intensity is unprecedented. Hyperscaler capex now reaches 45–57% of revenue at multiple firms — a level historically unthinkable for software-economics businesses. In 2024, the four hyperscalers combined spent approximately $230 billion. By 2026 the figure is ~$725 billion. No prior technology cycle, including the late-1990s telecom build-out, exhibited capital deployment at this absolute scale.

The supplier of choice has been re-rated. NVIDIA, founded in 1993 as a graphics processor company, crossed $1 trillion in market capitalization in 2023, $4 trillion in 2025, and $5 trillion shortly after. Its market value has multiplied more than twentyfold in five years. Long-term holders have realized returns exceeding 30,000% over a decade. NVIDIA today commands roughly 80% of the AI accelerator market.¹⁰

SECTION 3

Follow the Money: The Policy-Aligned Capital Stack

Hyperscaler capex is only part of the picture. Layered on top of it is a multi-year, policy-aligned capital stack of AI-tagged commitments — verified individual pledges from companies and sovereign partners — that together sum to approximately $5–6 trillion through 2028. The federal AI Action Plan published in July 2025¹¹ made the policy framework explicit: streamlined permitting for semiconductor fabrication facilities and the energy infrastructure required to power them, with the stated objective that the United States "build, baby, build" the AI stack ahead of geopolitical competitors. The capital is structurally committed to a five-year deployment horizon — meaning this cycle is not a one-year phenomenon subject to a single rate cycle or earnings season.

Has This Ever Happened Before? No.

Hyperscaler capex alone — at approximately 2.6% of U.S. GDP in 2026 — is already twice the peak share of the 1996–2001 telecom build-out, and more than three times the peak of the shale-energy boom. Adding the broader AI capital stack pushes the figure higher still. In the history of peacetime private investment, the United States has never seen this much capital committed to a single technology stack over a five-year window. Cycles end when capital flow stops; this capital flow is structurally committed and just beginning.

SECTION 4

Why This Cycle Is Different — And Larger

The most common objection to the AI thesis is that it resembles the dot-com bubble — a speculative wave of capital chasing companies without earnings, doomed to a similar drawdown. The data does not support this analogy. On the two dimensions that matter most — adoption velocity and underlying profitability — the AI cycle is structurally stronger than the internet cycle that preceded it.

Adoption Velocity: A New Category

ChatGPT reached 100 million users in approximately two months.¹² The World Wide Web took seven years. The mobile phone took sixteen. UBS analysts characterized the ramp as the fastest a consumer-internet application had ever achieved. The economic implication is direct: the time available for incumbent firms to defend their competitive positions is dramatically shorter than in any prior technology cycle, and the time available for capable firms to capture share is correspondingly larger.

Profitability: The Decisive Difference

At the dot-com peak in March 2000, the NASDAQ-100 forward P/E reached approximately 60×.¹³ Many of the era's most-followed names were unprofitable, and the average internet company lost 88% of its value in 2000. The AI cycle's leadership cohort looks fundamentally different: aggregate free cash flow margins above 20%, net income margins in the same zone, and forward P/E multiples in the high-20s to low-30s. These are not speculative ventures funded by IPO proceeds — they are among the most profitable enterprises in history, funding the AI build-out from operating cash flow.

SECTION 5

The Cost of Waiting

The 1995–2000 NASDAQ Composite, plotted alongside a representative basket of AI-cycle leaders from January 2023 to the present, shows that through their first three years, the two cycles have been substantially similar in shape, with the modern cycle slightly ahead. Whether the AI cycle continues to track, exceed, or diverge from the prior path is unknowable — but the historical analog itself is informative.

Investors who waited until 1999 to participate in the internet cycle entered after the bulk of the durable wealth had already been created in the surviving franchises. Those who participated from 1995 forward, with discipline, and stayed through the drawdown, owned Cisco, Microsoft, Amazon, and the predecessor of Google through the subsequent two decades. The early entrant who exercised discipline — not the speculator and not the skeptic — was the cycle's largest beneficiary.

The Lessons That Transfer

Survivors compound. Speculation does not. The dot-com cycle's casualty list (Pets.com, eToys, 360networks) is widely remembered. Less remembered is that the cycle's survivors — Amazon, Microsoft, Cisco, Apple — went on to produce some of the largest equity returns in market history. Capturing the survivors and avoiding the casualties is the central problem of cycle investing, and it is fundamentally a problem of factor selection.

Picks-and-shovels exposure has historically led the cycle. Cisco was the internet cycle's picks-and-shovels supplier. NVIDIA is the AI cycle's. The pattern of suppliers being re-rated first, while applied users are re-rated later as their use cases prove out, is highly consistent across cycles.

Industry productivity follows the build, with a lag. Goldman Sachs estimates AI could ultimately raise global GDP by approximately 7% (~$7 trillion) over a decade once adoption matures.¹⁴ McKinsey's higher-end estimates exceed $20 trillion annually by 2040.¹⁵ Even skeptics conceding modest near-term macro impact note that every prior general-purpose-technology cycle showed a productivity lag of approximately twenty years between breakthrough and broad measured impact. The early innings produce the equity returns; the later innings produce the GDP growth.

SECTION 6

WBI Power Factor AI™ — The Disciplined Path In

Recognizing the cycle is the easy part. Building disciplined exposure to it is harder — and the industry's product shelf for the AI investment opportunity remains remarkably thin. Most AI-themed ETFs are passive, market-cap-weighted, and apply no factor discipline; many are concentrated in a handful of mega-cap names that investors already own through their broad index exposure. WBI Power Factor AI™ was built to fill this gap.¹⁶

The POWER FACTOR AI Approach

WBI Power Factor AI screens an investable universe of approximately 190 AI-economy equities — spanning AI infrastructure, semiconductors, hyperscalers, AI software, and AI-applied platforms — and applies a multi-factor ranking engine that emphasizes quality, momentum, capture-ratio asymmetry, and downside risk control. The portfolio holds approximately twenty positions, weighted by quadratic rank decay with an 8% per-name cap. The approach is rules-based, point-in-time, and free of look-ahead bias.

Why Factor Selection Matters in a Thematic Cycle

Thematic exposure without factor discipline is exposure to the average — including the casualties. The dot-com cycle is the canonical illustration: holders of broad internet indices through 2000–2002 lost most of their gains, while disciplined factor selection captured the survivors. The AI cycle will produce its own casualties. The investor's task is not to avoid the cycle but to participate in it through a vehicle that systematically tilts toward the characteristics that differentiate the durable franchises from the also-rans.

For Advisors: Three Practical Considerations

Sizing. WBI Power Factor AI is designed as a thematic completion sleeve, sized as a satellite to a core allocation. Typical advisor implementations range from 5–15% of the equity allocation, depending on client risk profile and existing exposure to the Magnificent 7.

Rebalance discipline. The strategy rebalances on a defined cadence with a quadratic weighting framework that reduces position size after strength and increases it after weakness, in proportion to factor signal — capturing the cycle's volatility asymmetrically rather than chasing it.

Tax efficiency. For clients with concentrated single-stock AI positions — executives, founders, early employees — WBI Power Factor AI is the candidate strategy for a §351 ETF launch, enabling tax-deferred diversification of appreciated stock into a rules-based, factor-disciplined AI portfolio. Materials available on request.

SECTION 7

Risks and Honest Caveats

A definitive case for AI investment is not a case for unconditional optimism. Several risks deserve direct treatment, because advisors who present the opportunity without acknowledging them invite the very skepticism that erodes credibility.

Capex Returns Remain Unproven at Macro Scale

Goldman Sachs' chief economist publicly observed in early 2026 that the macro-level productivity contribution of AI investment in 2025 was "basically zero" — that is, while specific tasks have shown 30%+ productivity gains, the economy-wide effect remains undetectable in measured GDP.¹⁷ J.P. Morgan analysis suggests AI must generate $600 billion of annual revenue to justify a 10% return on installed infrastructure. Whether AI revenue catches up with AI capex is the central unresolved question of the cycle.

Concentration and Circular Financing

A meaningful share of recent AI capex commitments involves circular financing — infrastructure providers taking equity in AI customers who use the proceeds to buy back infrastructure. The reported NVIDIA–OpenAI memorandum of understanding is one example. These structures can amplify upside but also magnify downside if any node in the network falters.

Valuation Is Not Cheap

The Shiller CAPE ratio has crossed the 40 threshold¹⁸ — a level seen previously only at the dot-com peak. Forward P/E multiples for the AI cohort, while less stretched than in 2000, are well above long-term averages. A market correction of 20–30% would not invalidate the AI thesis, but it would test investor discipline. Position sizing and risk management matter accordingly.

Regulatory and Geopolitical Overhang

Export controls on advanced semiconductors, antitrust scrutiny of large platforms, and competition from Chinese AI providers (DeepSeek demonstrated the cost-equivalence of frontier capability earlier than markets expected) all represent material risks to the leadership cohort.

"None of these risks alters the underlying thesis that AI is a multi-decade wealth-creation event. They reframe how to participate: with discipline, with appropriate sizing, with factor selection that adapts as the cycle evolves, and with explicit risk management."

Document version:

WBI WealthCreation White Paper.pdf

Download PDF • 241KB

Disclosures & Sources

*For institutional use only.

Past performance is not indicative of future results. This is not an offer to buy or sell any security. No security, including those referred to directly or indirectly, is suitable for all accounts or profitable all the time. This information is compiled from sources believed to be reliable, but accuracy cannot be guaranteed. You should not assume that any discussion or information provided here serves as a substitute for personalized investment advice from WBI or any other investment professional. If you have any questions regarding the applicability of specific issues discussed to your individual situation, please consult with WBI or your chosen investment advisor. Additional information about WBI’s advisory operations, services, conflicts of interest and fees are in the Form ADV, which is available upon request or on the SEC’s website at http://www.adviserinfo.sec.gov. WBI is a registered investment adviser. Registration of an Investment Adviser does not imply any level of skill or training.

Net of Fee Performance is net of WBI’s maximum investment management fees. This model fee approach consists of netting down 100 bps from gross returns on a monthly basis. The actual, annual investment fee rate charges shall vary typically between 75 bps and 100 bps, but not more than 100 bps) depending upon market value of assets under management and the specific type of investment services to be rendered.

Other strategies may have different results.

References to other securities is not an offer to buy or sell.

© 2026 WBI Investments, LLC. All Rights Reserved.

Forward-Looking Statements. This document contains forward-looking statements regarding artificial intelligence adoption, capital expenditure cycles, and potential investment outcomes. Forward-looking statements are inherently uncertain, depend on assumptions that may not prove correct, and are not guarantees of future results. Actual outcomes may differ materially from those discussed.

References to historical wealth-creation events are illustrative analogies, not predictions. Charts comparing prior cycles to the current AI cycle are constructed from approximate index and basket data and are intended to illustrate structural parallels, not to forecast specific returns.

WBI Power Factor AI™ is a proprietary investment strategy of WBI Investments. Strategy descriptions herein are summaries; full methodology and performance disclosures are available in separate strategy documentation. Investors should review the relevant strategy document and consult their investment advisor before making any allocation decision. Power Factor AI™ is a trademark of WBI Investment Management.

Risks of Thematic and Sector-Concentrated Strategies. Thematic strategies concentrate exposure in companies sharing a defined characteristic. Such strategies are subject to heightened risks of single-theme reversal, regulatory action, valuation compression, and capital cycle correction. Investors may lose principal.

Key Data Sources:

Hyperscaler capital expenditure: company quarterly earnings reports and guidance (Microsoft, Alphabet, Amazon, Meta, Oracle); Financial Times compilation, CNBC, CreditSights, Futurum Group analyses, Q4 2025 / Q1 2026. ft.com | cnbc.com

NVIDIA market capitalization: CNBC, Morningstar. Adoption velocity: UBS, Statista, Epoch AI. statista.com

Productivity and GDP estimates: Goldman Sachs Global Investment Research; McKinsey & Company; MIT Sloan / Acemoglu, "The Simple Macroeconomics of AI" (2024). goldmansachs.com | mckinsey.com

Dot-com cycle metrics: NASDAQ historical data; BIS analysis. CAPE ratio: Shiller / Yale. econ.yale.edu/~shiller

Footnotes:

1. Hyperscaler capex figures: Microsoft, Alphabet, Amazon, Meta quarterly earnings reports (Q4 2025 / Q1 2026); Financial Times and CNBC compilations. See: ft.com and CreditSights, Futurum Group analyses.

2. NVIDIA market capitalization history: CNBC, Morningstar. See: cnbc.com

3. ChatGPT adoption velocity: UBS, Statista, Epoch AI. Figures are widely-cited approximations. See: statista.com

4. Hyperscaler capex figures: Microsoft, Alphabet, Amazon, Meta quarterly earnings reports (Q4 2025 / Q1 2026); Financial Times and CNBC compilations. See: ft.com and CreditSights, Futurum Group analyses.

5. NVIDIA market capitalization history: CNBC, Morningstar. See: cnbc.com

6. ChatGPT adoption velocity: UBS, Statista, Epoch AI. Figures are widely-cited approximations. See: statista.com

7. WBI Power Factor AI™ is a proprietary strategy of WBI Investment Management. Full methodology and performance disclosures available in separate strategy documentation. Contact WBI for details: wbiinvestments.com

8. Hyperscaler capex figures: Microsoft, Alphabet, Amazon, Meta quarterly earnings reports (Q4 2025 / Q1 2026); Financial Times and CNBC compilations. See: ft.com and CreditSights, Futurum Group analyses.

9. Hyperscaler capex figures: Microsoft, Alphabet, Amazon, Meta quarterly earnings reports (Q4 2025 / Q1 2026); Financial Times and CNBC compilations. See: ft.com and CreditSights, Futurum Group analyses.

10. NVIDIA market capitalization history: CNBC, Morningstar. See: cnbc.com

11. Federal AI Action Plan (July 2025): The White House Office of Science and Technology Policy. See: whitehouse.gov/ostp

12. ChatGPT adoption velocity: UBS, Statista, Epoch AI. Figures are widely-cited approximations. See: statista.com

13. Dot-com cycle metrics: NASDAQ historical data, Bank for International Settlements analysis. Wikipedia compilation. See: bis.org

14. Goldman Sachs Global Investment Research, "Generative AI Could Raise Global GDP by 7%" (2023); subsequent 2025–2026 updates. See: goldmansachs.com

15. McKinsey & Company, "The Economic Potential of Generative AI" (2023). See: mckinsey.com

16. WBI Power Factor AI™ is a proprietary strategy of WBI Investment Management. Full methodology and performance disclosures available in separate strategy documentation. Contact WBI for details: wbiinvestments.com

17. Goldman Sachs Global Investment Research, "Generative AI Could Raise Global GDP by 7%" (2023); subsequent 2025–2026 updates. See: goldmansachs.com

18. Shiller CAPE ratio: Yale/Shiller data. See: econ.yale.edu/~shiller

19. WBI Power Factor AI™ is a proprietary strategy of WBI Investment Management. Full methodology and performance disclosures available in separate strategy documentation. Contact WBI for details: wbiinvestments.com