-Insights-
Not All Risk Is Created Equal
Diversification is often evaluated by counting what a portfolio owns.
An investor may hold dozens of companies, several exchange-traded funds, and exposure to nearly every major market sector. On a conventional allocation report, the portfolio may appear broadly diversified.
But the number of holdings does not necessarily reveal the number of independent risks.
A portfolio can own technology companies, utilities, industrial manufacturers, real estate operators, and financial institutions while remaining heavily dependent on one economic narrative. Today, the rapid expansion of artificial-intelligence infrastructure provides a particularly useful example.
What began as an investment story centered on software developers and advanced semiconductors has spread throughout the economy. Building and operating AI systems requires processors, memory, networking equipment, data centers, cooling systems, electrical infrastructure, power generation, construction financing, and substantial amounts of capital.
The companies supplying those needs come from different industries. Their fortunes, however, may increasingly depend on the same assumption: that AI demand will grow quickly enough – and generate enough economic value – to justify an extraordinary infrastructure buildout.
The portfolio may contain many sectors. It may still be concentrated in one story.
From a technology trend to a capital-investment cycle
The most visible participants in the AI theme have been the major technology platforms and semiconductor companies. But an AI data center cannot operate on processing chips alone.
It also requires specialized memory to move large quantities of data, networking equipment to connect thousands of processors, cooling systems to manage heat, and reliable access to large amounts of electricity. Those needs draw industrial companies, utilities, energy producers, and data-center developers into the same ecosystem.
They also require financing
The scale of expected investment means that operating cash flow alone may not fund the entire buildout. Research from the Bank for International Settlements indicates that anticipated AI investment needs are likely to push companies toward greater use of debt, with private credit playing a growing role. The BIS also describes off-balance-sheet arrangements that can function economically like borrowing even when the obligations do not appear as conventional corporate debt.
That brings banks, insurers, asset managers, private-credit funds, bond investors, and structured-finance markets into the AI investment chain.
A hypothetical portfolio could therefore own a cloud-computing company, a semiconductor manufacturer, a memory producer, an electrical-equipment supplier, a utility, a data-center real estate company, and a financial institution involved in funding data-center construction. The businesses look different. The dominant economic driver may not be.
The circular nature of the ecosystem
A technology company forecasts strong demand for AI computing and commits to additional infrastructure. A data-center developer uses that commitment – or a long-term lease from the technology company – to obtain construction financing. Banks, private-credit managers, or bond investors provide the capital. The developer uses the proceeds to purchase chips, cooling equipment, and electrical systems. Those orders support supplier revenue, encouraging suppliers to add capacity of their own.
Some of those suppliers may also borrow to build factories, expand production, or secure long-term inputs. The resulting activity supports economic growth, employment, and earnings throughout the ecosystem. Strong reported growth can then reinforce investor confidence in the original AI demand thesis.
This does not make the activity artificial or inherently unsustainable. Much of the spending is financing real physical assets backed by companies with substantial revenues and strong balance sheets.
The structure can create circular dependencies.
The creditworthiness of a data-center project may depend on the long-term commitment of a major technology tenant. That tenant’s willingness to honor or expand the commitment may depend on future demand for AI services. The economics of those services may depend on the availability and cost of computing capacity. Meanwhile, the suppliers producing that capacity may be investing based on orders generated by the same expected demand.
Capital travels around the ecosystem, but the ultimate source of repayment remains the economic value produced by AI applications.
If that value develops as expected, the funding chain can support a lasting expansion in productivity and infrastructure. If monetization arrives more slowly, several apparently separate investments could face pressure at the same time.
Why financials belong in the concentration discussion
Financial companies do not manufacture semiconductors or operate data centers, but they can still become exposed to the AI investment cycle.
Banks may provide construction loans, revolving credit facilities, warehouse financing, or funding lines to private-credit vehicles. Investment banks may underwrite corporate bonds, project debt, asset-backed securities, or commercial mortgage transactions. Insurers and asset managers may ultimately hold portions of the debt. Private-credit funds may finance projects that are too specialized, early-stage, or complex for conventional public markets.
The Federal Reserve Bank of Dallas has noted that AI-related financing could increase the supply of long-duration corporate debt and influence broader interest-rate markets. Funding can move through investment-grade bonds, private-credit loans, and transactions that convert floating-rate exposure into longer-duration securities.
Structured-finance markets are also becoming more involved. Fitch Ratings has documented rapid growth and evolving structures in data-center asset-backed securities, commercial mortgage-backed securities, and project finance. These developments do not establish that banks or credit investors face an imminent problem. They do show that AI exposure can migrate beyond equity markets.
A slowdown in AI infrastructure spending could therefore affect more than technology earnings. It could influence loan demand, underwriting activity, bond issuance, credit spreads, private-asset valuations, and the performance of securities backed by data-center projects.
Different sectors, shared assumptions
Technology companies may depend on rising demand for computing services. Semiconductor and memory companies may depend on continued equipment orders and favorable pricing. Utilities may depend on projected data-center loads materializing. Industrial companies may expand production based on anticipated construction. Real estate developers may rely on leases from a concentrated group of tenants. Financial institutions and credit investors may underwrite debt based on those contracts and growth assumptions.
The risks are not identical. A utility faces different regulations than a semiconductor company. A secured data-center loan has a different capital structure than common equity in a cloud provider. A cooling-equipment manufacturer may have different margins and competitive dynamics than a memory producer.
But diversification is not simply a matter of confirming that the holdings are different. It requires understanding which assumptions they share.
The AI theme could continue growing while individual investments disappoint. A shortage may become an oversupply as new capacity comes online. A critical bottleneck may move from chips to memory, power, networking, or financing. A company may report strong growth while its stock falls because expectations had risen even faster. A data-center project may be economically viable but still suffer delays, cost overruns, or changes in financing terms.
Market leadership can rotate without invalidating the long-term theme. At the same time, a strong long-term theme does not eliminate valuation, execution, balance-sheet, or credit risk.
Diversifying the story, not merely the holdings
Traditional portfolio analysis usually identifies concentration by company, industry, sector, or asset class. Those measures remain important. But the spread of AI across the economy illustrates why advisors may also need to consider economic-driver concentration.
That means asking what underlying development is expected to produce the return and what could cause several holdings to disappoint together.
How much of the portfolio relies on continued hyperscaler spending? How much depends on AI services becoming profitable? How much depends on uninterrupted access to affordable power and financing? Are the financial exposures tied to a broad range of borrowers, or concentrated among projects serving the same small group of tenants? Would the holdings still diversify one another if credit conditions tightened or infrastructure plans were delayed?
The objective is not to avoid a promising economic trend. It is to separate confidence in that trend from the amount and form of risk the portfolio should carry.
Beacon’s broader investment framework recognizes that risk can emerge at several levels: across the broad market, within individual holdings or sleeves, and through changes in the economic environment. No single lens captures every way a portfolio can become vulnerable.
A portfolio can own many companies and still depend on very few outcomes.
The new concentration risk is not merely owning too much of one stock or sector. It is owning several different investments that all require the same story to remain true.
Bank for International Settlements. Financing the AI Boom: From Cash Flows to Debt. BIS Bulletin No. 120, 2026.
https://www.bis.org/publ/bisbull120.htm
Bank for International Settlements. Financing the AI Infrastructure Boom: On- and Off-Balance-Sheet Borrowing. BIS Quarterly Review, March 2026.
https://www.bis.org/publ/qtrpdf/r_qt2603u.htm
Federal Reserve Bank of Dallas. How AI Debt Financing Impacts Duration Supply and Interest Rates. February 10, 2026.
https://www.dallasfed.org/research/economics/2026/0210-searls-aifinancing
Fitch Ratings. Data Center Evolution Highlights Structured and Project Finance Risk Considerations; and Global Structured Finance and Project Finance Data Center Ratings Market Update and Q&A. July 13, 2026
https://www.fitchratings.com/research/structured-finance/global-structured-finance-project-finance-data-center-ratings-market-update-q-a-13-07-2026
The views and opinions expressed are my views and opinions as an individual and do not reflect the views and opinions of Beacon Capital Management, Inc.
Beacon Capital Management, Inc. is a registered investment adviser. Information presented herein is for educational purposes only. Beacon Capital Management does not provide tax advice and strongly urges that retail investors consult with their tax professionals regarding any potential investment.
JSO-080326-A