Insights

TrueNorth: Navigating a Boom, a Shock, and Life-Changing Technology

NorthRock Investment Team
•
November 13, 2026

Key Takeaways

A quick look at where markets have been, and what we believe lies ahead.

  • A dollar put into US stocks 15 years ago is worth roughly eight today, according to the S&P 500 total return index.1 We believe the next few years are unlikely to be as easy.
  • Shocks pass; booms end. The pandemic and the Iran war knocked the economy off course for a while. We believe shocks tend to pass, while booms can end painfully. In our view, the AI spending boom is concentrating a great deal of capital in one place, which creates risk for investors.
  • Business spending on data-center computers is up 223% since ChatGPT arrived in late 2022.2 Investors have piled into the hardware, just as they piled into internet equipment makers in the late 1990s. Some of those makers failed, and some survivors were still underwater 20 years later.
  • We estimate that the four largest AI spenders will spend more than $1 trillion in 2027.3 We are not calling the end of the boom in 2027, but chip and hardware companies have recently been nearly 30% of the S&P 500’s market value,4 so we watch that exposure closely.
  • We believe the long-term beneficiaries of AI may include innovators creating new products and services and cybersecurity firms.5
  • Bonds pay again: the 10-year Treasury yields nearly ten times its 2020 low.6 Higher yields mean bonds offer more income than they did in 2020. We believe higher-quality bonds may help cushion portfolios in some equity downturns.7

Equity market returns have come “easy” since 2010. That’s when the full extent of the historic financial crisis was felt, and the US and world economy began to recover. Since then, US equities have suffered just two bear markets, using the S&P 500 total return index as our basis. During Covid, the S&P fell 34% during the month of the pandemic’s onset in March 2020. But because of the government’s interventions, US shares recovered all their losses in a mere five months. Fears that the US economy would enter a recession after the tariff announcements in April 2025 were even shorter-lived.

We believe the market environment ahead may be more challenging than the past 15 years.8 The war in Iran and the change in Fed leadership have caused interest rates to go higher. In addition, the AI boom in spending is distorting equity markets and valuations, even as major hyperscalers9 plan to spend even more in the years to come.

At NorthRock, we believe it is important to distinguish between temporary, “exogenous” shocks that briefly divert the economy and “booms” like AI that often misalign resources leading to major changes in equity indices when they end. The pandemic of 2020 and the war in Iran this year were two very different types of shocks. We believe energy may be more costly and global trade less secure as a result of the Iran war. The eventual end of the AI boom could be more disruptive for markets, though timing and impact are uncertain.

Where We Are, Where We Are Going

Over the past 15 years, US equity returns have compounded at about a 15% annualized rate, well above the long-term average of 10%.10 Put differently, a dollar invested in the S&P 500 fifteen years ago is worth roughly eight dollars today with dividends reinvested. The past decade alone has delivered returns of about 15% a year. High-quality US bonds returned less than 2% a year over the same fifteen years, and developed international shares roughly 8%.

For a decade and a half, US stocks have outpaced the other major asset classes we track,11 and many investors have known little else. What’s helped of late is a surge in technology spending for AI development. Business investment in the computers that fill data centers has surged 223% since ChatGPT dazzled the world with its “large language (AI) model” in late 2022.12

Today, we may be on the cusp of seeing the productivity benefits of a technology so powerful that it could augment or replace human labor. The advent of “AI agents” joining the workforce looks like an event on the scale of the Industrial Revolution. Even so, we doubt that it has changed the remarkably consistent pattern in financial markets we’ve experienced around past technological breakthroughs such as the locomotive or the internet.

While we are optimists on the impact of the technology on the economy over the long term, the foundational elements of AI’s early development present true “boom and bust” risks for the economy and the stock market.

Building the infrastructure to run AI won’t continue at the same pace indefinitely. For illustration, if capital spending at the four largest AI spenders kept growing at its 2023-2026 pace, it would exceed the size of the US economy by 2035.13 Business investment in information processing equipment has already risen from 1.8% of GDP in late 2022 to 2.8%, close to its dot-com peak of 2.9% in late 2000.14

Yet, over the past two years, investors have followed a familiar course. They have focused less on the corporate beneficiaries of AI services, or even on those who will provide the AI services themselves, than on the hardware on which AI runs. This is most reminiscent of the late 1990s, when investors poured their capital into makers of the communications equipment that enabled internet services. While web content flourished after the bust in equipment and transmission (fiber optic) spending, some of the largest firms that provided the internet gear failed to survive. The shares of a few that did suffered a negative return over the subsequent twenty years.15

The market may eventually shift from rewarding hardware providers toward rewarding companies that use AI, though timing is uncertain and today’s leaders may continue to perform well. As with the internet before it, running AI software and services is distinct from building AI’s “pipes.”

Figure 1: IT Equipment Investment and Semiconductor and Hardware Share of S&P 500 Market Cap, 1990 to 202616

Line chart, 1990 to 2026: IT equipment investment as a percent of GDP and the semiconductor and hardware share of S&P 500 market cap. Both peaked around 2000, fell, and have risen sharply since 2023 to near or above their dot-com highs.

Of course, there are differences between the current tech boom and the last one, in the 1990s. In our view, the biggest spenders are strong, mature and diversified companies, not “experiments.” We believe demand for computing power is still running ahead of capacity. Most clearly, we estimate that spending at the four largest AI spenders will rise 48% in 2027, following 89% in 2026.17 On our estimate, that takes their spending past $1 trillion in 2027.

With this in mind, we are not calling an end to the boom this coming year or even an end to the outperformance of shares of companies that are presently profiting from the boom. Yet, unsustainable rates of growth cannot endure. Strains on resources, including debt capital to finance the buildout, suggest we need to carefully manage exposure, including record concentration risk in semiconductor shares.

Not 30% Forever

Semiconductor and tech hardware shares have recently represented nearly 30% of S&P 500 market capitalization. That looks like investors betting that tech hardware firms will earn roughly that share of all future US corporate profits. Yet the one-time buildout that drove semiconductor shares may not repeat, and maintaining infrastructure may not be enough to support the same pace of growth.

We believe equity capital should always focus on innovators driving growth and creating new demand in the economy. However, investing in commodity memory-chip producers can be very different from investing in the emerging leaders in AI services or the cybersecurity software needed to defend against its darker capabilities, and different sectors can lead at different times. What we call Active Asset Allocation is distinct from an approach that is static.18

We also believe the period ahead carries more geopolitical risk than the one behind us. The political backdrop, at home and abroad, looks less predictable than at any point in the post-2010 recovery; federal deficits are running near $2 trillion a year, roughly 6% of GDP, in an economy at full employment, with gross federal debt now larger than the economy itself;19 and AI’s effect on employment, beginning with entry-level and routine white-collar work, could weigh on incomes and confidence even as it lifts productivity.

Therefore, with the AI buildout that has powered markets looking likely to slow and other risks rising, we want to prepare portfolios for a range of outcomes, not a single call.

What Assets Absorb Risk?

The rapid growth in borrowing for AI may well continue in 2027. Yet it still represents a small slice of overall corporate debt.20 US government borrowing has grown alongside the AI boom, an unusual situation that has caused upward pressure on bond yields and losses for many bond prices in 2026. But what is striking is how far bond yields have risen since reaching history’s lowest levels in 2020-2021. Today, the yield to maturity on a 10-year US Treasury is nearly 10X the low of 2020.21

Investors were right to avoid large bond allocations – acting as lenders – during the years of “zero yields.” By definition, a “static” 60/40 asset allocation does not change even when the prices of bonds surge to what we see as dangerous levels, as they did before history’s largest loss for the Bloomberg US Aggregate Bond Index in 2022.22

In contrast, our Active Asset Allocation methodology seeks not to extrapolate the past forward. By taking into account today’s valuation for each investment, it seeks not to assume a repeat of the past.23 For example, at today’s yields, higher-quality bonds may help absorb some risk in a future equity bear market.24

It is our job to be forward looking and seek opportunities where we believe future profits may grow faster than the economy as a whole.25 It is similarly our role to build more resilient portfolios as circumstances change and “boom” risks emerge.

Disclosures

This material is for educational purposes only and is not personalized investment, tax, or legal advice, or a recommendation to buy or sell any security or take any particular action. Views expressed are the authors’ opinions as of November 9, 2026, may change without notice, and are not guarantees of future results. Forward-looking statements, including market outlooks and economic projections, may not come true. Statements about individuals and firms reflect the authors’ opinions.

NorthRock Partners LLC has a strategic partnership with CIO Capital Group, LLC (“CIOCG”). NorthRock receives a share of the revenue from CIOCG strategies in which client assets may be invested, has a cross-referral arrangement with CIOCG, and includes CIOCG personnel, including David Bailin and Steven Wieting, on its Investment Committee. These arrangements give NorthRock a financial incentive to recommend CIOCG, which is a conflict of interest. Mr. Bailin and Mr. Wieting are CIOCG principals, not NorthRock employees or clients, and their statements about NorthRock reflect their own opinions.

Investing involves risk, including loss of principal. Risk management, asset allocation, and diversification do not guarantee profits or prevent losses. Active strategies may underperform benchmarks. Past performance does not predict future results. Indexes are unmanaged, do not reflect fees, and cannot be invested in directly; index and asset-class returns shown are for illustration and are not NorthRock or client performance. Bond prices fall when interest rates rise, and bonds, Treasuries, and gold can lose value, including during equity market declines. Investments concentrated in a sector, such as technology or semiconductors, involve additional risk. Sectors and themes discussed are the authors’ views, are not recommendations, and may not be reflected in any client’s portfolio.

Portfolio implementation varies by account; changes to non-discretionary accounts require client approval. Statistics and data are from the sources and dates cited, are believed reliable, and have not been verified.

All investment advisory and Personal Office® services are provided by and through NorthRock Partners LLC, an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply any particular level of skill or training. See NorthRock’s Form ADV Part 2A and Wrap Fee Program Brochure for more information about conflicts, fees, and risks.

Sources

  1. Source: Bloomberg, S&P 500 total return index (gross dividends), month-end October 2011 to month-end October 2026: $1 grew to $7.93, about 14.8% a year.
  2. Source: US Bureau of Economic Analysis via Haver Analytics. The GDP accounts do not report data-center spending separately. The figure is private investment in computers and peripheral equipment, the closest line item, in nominal terms at a seasonally adjusted annual rate: $155.9 billion in the fourth quarter of 2022 and $504.1 billion in the third quarter of 2026, the latest available and subject to revision. Adjusted for inflation, the increase is 171%.
  3. CIO Group estimate as of November 9, 2026. Company guidance for 2026 totals about $720 billion to $740 billion for the four companies, on mixed definitions as of October 30, 2026. Each company says 2027 spending will be higher, but none gives a figure. Our 2026 and 2027 estimates apply growth of 89% and 48% to a 2025 base of $376 billion.
  4. Source: S&P Dow Jones Indices via Haver Analytics. Market capitalization of S&P 500 Semiconductors and Semiconductor Equipment plus Technology Hardware and Equipment, as a share of S&P 500 market capitalization: 29.9% at the end of June 2026, 29.8% in September and 27.2% in October.
  5. Sector investing carries risk, and past patterns may not repeat.
  6. Source: US Treasury 10-year constant-maturity yield via Haver Analytics, daily: 4.88% on November 9, 2026 and 0.52% on August 4, 2020.
  7. Bonds can also lose value, including when interest rates rise.
  8. Returns are uncertain and could be better or worse than we expect.
  9. Hyperscalers are the very large cloud-computing companies that build and run huge data centers.
  10. Source: Bloomberg and Damodaran (NYU Stern). S&P 500 total return index, month-end October 2011 to October 2026: 14.8% a year, so $1 became $7.93; the ten years to October 2026: 15.4% a year. Over the same fifteen years, Bloomberg US Aggregate Bond Index: 1.9% a year; MSCI EAFE Index, net, in US dollars: 7.8% a year. Long-term average: S&P 500 including dividends, 1928 to 2024, 9.9% a year (Damodaran, NYU Stern).
  11. Annualized total returns, October 2011 to October 2026: S&P 500 14.8%; S&P Listed Private Equity Index 11.5%; US equity REITs (FNERTR) 7.9%; MSCI EAFE Index, net, in US dollars, 7.8%; gold (spot price) 6.0%; Bloomberg US Aggregate Bond Index 1.9%; Bloomberg US Treasury Long Index 0.8%. Source: Bloomberg. The listed private equity, REIT, gold and long Treasury figures run to October 19-20, 2026, the latest available.
  12. Private investment in computers and peripheral equipment, the closest line in the GDP accounts to data-center spending; nominal, fourth quarter of 2022 to third quarter of 2026. Source: US Bureau of Economic Analysis via Haver Analytics. See also note 2.
  13. Capital spending of the four largest AI spenders: $147 billion in 2023 and $695 billion in 2026 (reported through 2025, consensus for 2026; Bloomberg, supplied mid-September 2026), or about 68% a year. Extended at that pace it passes nominal US GDP in 2035, with GDP of $32.5 trillion in 2026 growing 4% to 5.3% a year (Haver Analytics). The result depends on the start year: from 2022 the pace is about 46% a year and the crossing moves to about 2038. On the BEA’s narrower series for business investment in computers and peripheral equipment (44% a year since 2023), the crossing is 2040.
  14. Private nonresidential fixed investment in information processing equipment as a share of GDP, both at seasonally adjusted annual rates: 1.8% in the fourth quarter of 2022, 2.8% in the third quarter of 2026 and 2.9% in the fourth quarter of 2000. Source: US Bureau of Economic Analysis via Haver Analytics.
  15. Examples, total return with dividends reinvested, March 31, 2000 to March 31, 2020: Cisco -34%, Corning -58%, Ciena -91%, JDS Uniphase (now Viavi) -98%. Nortel filed for bankruptcy protection in January 2009, and Lucent was acquired by Alcatel in 2006. Source: Financial Modeling Prep, dividend-adjusted prices.
  16. Source: CIO Group, Haver Analytics, as of October 2026. IT equipment investment is private nonresidential fixed investment in information processing equipment (US Bureau of Economic Analysis), quarterly through the third quarter of 2026, as a share of GDP. The semiconductor and hardware share is the S&P 500 market capitalization of semiconductors and semiconductor equipment plus technology hardware and equipment, month-end through October 2026, as a share of S&P 500 market capitalization. Past patterns may not repeat and do not predict future results.
  17. CIO Group estimate as of November 9, 2026. Company guidance for 2026 totals about $720 billion to $740 billion for the four companies, on mixed definitions as of October 30, 2026. Each company says 2027 spending will be higher, but none gives a figure. Our 2026 and 2027 estimates apply growth of 89% and 48% to a 2025 base of $376 billion.
  18. Active Asset Allocation is an approach that seeks to adjust portfolios as valuations and risks change.
  19. Sources: Congressional Budget Office, Office of Management and Budget, US Treasury and Bureau of Labor Statistics, via Haver Analytics. Fiscal 2026 figures are budget estimates, not outcomes: the deficit is 6.4% of GDP, about $1.85 trillion (CBO) to $2.06 trillion (OMB), after 5.9% in fiscal 2025. Gross federal debt was 123% of GDP in fiscal 2025; debt held by the public was 99.4% in fiscal 2025 and 100.6% in the fiscal 2026 estimate. Unemployment was 4.1% in October 2026.
  20. Barclays Private Bank, October 9, 2026: AI-related debt is about 7% of the US investment-grade index and could reach 12% to 15%. Our register of AI-related new borrowing by public filers, chipmakers excluded: $149 billion in 2025 and $361 billion through November 4, 2026. Published 2027 forecasts for hyperscaler debt range from $220 billion to $300 billion (JPMorgan Asset Management, after $279 billion in 2026) to about $420 billion (Goldman Sachs, from about $250 billion in 2026).
  21. 10-year Treasury constant-maturity yield, daily, via Haver Analytics: 4.88% on November 9, 2026 and 0.52% on August 4, 2020 (see note 6). Yields change daily.
  22. Source: Bloomberg US Aggregate Bond Index, calendar-year total return: down 13.0% in 2022.
  23. No methodology can guarantee results.
  24. Bonds can lose value, including when interest rates rise, as in 2022, when the Bloomberg US Aggregate Bond Index fell 13.0% and the S&P 500 total return index fell 18.1%, calendar-year total returns, Bloomberg. Peak to trough in each S&P 500 decline (March 24, 2000 to October 9, 2002; October 9, 2007 to March 9, 2009; January 3 to October 12, 2022), the US Aggregate returned +29.1%, +7.2% and -14.4%, while the S&P 500 fell 24.5% in 2022 (total return). Source: Bloomberg.
  25. Our views may be wrong.