Previewing the Year-End Slipstream

We remain optimistic about AI’s rollout and are focused across three interrelated dimensions: supply, demand and regulatory. Supply remains a principal issue, with NVIDIA CEO Jensen Huang noting in the company’s latest earnings report that shortages exist across several phases, including high-speed memory chips, silicon vendors and AI-focused data centers.² In the past week alone, additional earnings reports from Broadcom, Dell and Hewlett Packard Enterprise also emphasized memory and data-center buildouts as primary growth constraints.^3,4,5

From a demand perspective, Huang emphasized during last week’s earnings report that although they were comfortable guiding investors to 70% revenue growth for its next fiscal year, that amount falls short of where they saw total demand across their client verticals.⁶ This week, Broadcom CEO Hock Tan noted that demand trends are extending across the AI stack, forecasting that AI networking revenue will grow just as fast over the next few years as specialized semiconductors have in more recent years.⁷

On the regulatory front, this week’s Group of 20 (G20) Innovation Ministerial in Chapel Hill, North Carolina included SpaceX and Tesla CEO Elon Musk, Meta CEO Mark Zuckerberg, Palantir CEO Alex Karp, Anthropic co-founder Tom Brown, OpenAI CEO Sam Altman and Huang. These leaders joined policymakers from around the world and other industry magnates in forming the Carolina Principles for Emerging Technologies, a consensus statement balancing technological growth with data and enterprise security.⁸

Despite a seemingly unified approach, how individual countries or regions respond to inevitable mishaps within this burgeoning technological landscape will shape growth prospects. While this medium does not offer a complete analysis of each dimension, we continue to see AI’s uptake and diffusion across an increasing number of users and its supply dynamics remaining in a balanced equilibrium. However, using history as a guide, that equilibrium can be upended by exuberance among buyers or sellers or by misalignment on policy oversight, so careful analysis remains paramount.

Capital markets offer a real-time barometer on AI-related spending. We have seen several instances over the past two earnings cycles where companies have increased their expenditures relative to prior guidance, and the response has been nearly binary: companies that increase spending while monetizing current opportunities (Amazon, Microsoft) have been rewarded, while companies increasing AI-related spending with less monetization emphasis have been punished (Meta, Oracle, Tesla). As noted above, the supply/demand imbalances we observe suggest that more spending growth will likely ensue, but what we are most focused on is the rate of change within capital budgeting plans and what marginal changes may accelerate or decelerate current projections.

Bloomberg analysts offer a helpful forecast of three major AI suppliers: Tier 1 hyperscalers (Amazon, Alphabet, Meta, Microsoft, Oracle), Tier 2 hyperscalers (next level down in total spend) and Chinese hyperscalers seeking to keep up China’s participation relative to global peers. In their most recent update in late August, Bloomberg anticipates that these aggregated groups will spend $976 billion this calendar year, a 95% increase from last year.⁹ For the next two calendar years, analysts expect $1.3 trillion and $1.5 trillion, large numbers but growth rates markedly lower from the prior three years where annualized growth averaged in the 80% range.¹⁰ Those numbers could certainly be revised higher once company guidance solidifies in current quarters, and these estimates reflect just a sampling of the broader AI ecosystem and exclude spending plans for memory and other hardware providers.

Our working thesis is that markets envision a more glidepath-driven deceleration in spending trends, even as adoption and usage rates grow. Stanford University’s 2026 Artificial Intelligence Report posits that generative AI usage in the United States is just 28% at the consumer level.¹¹ Further, AI business adoption seems to follow a “deep but shallow” pathway, with 70 to 75% of businesses using some form of AI, but experimentation and narrow use case adoption being the norm with current commercial users.¹² Back to the equilibrium framing, a glidepath lower on spending growth rates, monetization and return on capital discipline plus increased uptake are all critical variables for our ongoing assessment.

As we covered at length last week, we viewed communication between the Fed and capital markets as having taken a significant step forward during Chairman Kevin Warsh’s Jackson Hole Symposium keynote speech. While Chair Warsh continues to distance himself from providing “forward guidance” or foretelling intended central bank actions, he did offer insights into his assessment of current conditions. In addition, Warsh was clear that inflation is the Federal Reserve’s principal objective and that he would not waver from the Fed’s stated 2% inflation objective.

The big reveal for the rest of the year will be details around what Warsh outlined in June during his first press conference following the Federal Reserve’s Open Market Committee meeting. Warsh revealed that he had formed five task forces with the goal of advancing monetary policy. These task forces include Fed communications, how the Fed uses its balance sheet and the current balance sheet regime amassed during the 2008-09 Financial Crisis and again during the COVID-19 pandemic, what data the Fed uses to inform their decisions, how technology is impacting the labor market, and how the Fed understands and responds to inflation.¹⁵

From a capital markets perspective, we divide up the impacts from each task force into nearer- and longer-term implications. Nearer-term, the data and communications task forces may provide hints into what variables the Fed may be more focused on relative to past practices, as well as any changes to the information transfer mechanism investors have grown accustomed to. Longer-term, the Fed’s balance sheet usage, how they are interpreting AI’s economic impact, and their inflation framing may offer more structural insights. If we were to pick one of most import, changes to the balance sheet would receive top billing given its sheer size, and a departure from past practice would require a market adjustment.

The global interest rate backdrop has had our attention for some time, and frequent readers of this publication may have already grown weary of our oft-used phrase “the bond market remains in control of capital market outcomes.” While we look across maturities, borrowers and geographies, sovereign or government bonds are our primary focus right now. Governments are unique borrowers since they collect taxes and have multiple potential revenue sources, offering investors seemingly “risk-free” returns as long as governments remain stable and borrowings do not become overextended. Further, while so much focus remains on how central banks may alter interest rate policy, we find longer-maturity bonds more instructive when assessing the current capital market zeitgeist, including how auctions perform and how nuanced instruments like swaps perform.

Looking across global sovereign bonds, virtually every bond maturing 10 years from today is close to or at its highest yield and lowest price in the past 12 months, with exceptions including Switzerland, China and Argentina for more country-specific reasons. The balance of major borrowers are within 0.1% or 10 basis points of their yields’ respective apexes. This is driving some market participants to question if the confluence of increased issuance, growing indebtedness and seemingly embedded inflation means we are entering a new borrowing regime for governments and, by implication, consumers and businesses who do not have taxing power and whose borrowing costs take their cue from sovereigns.¹⁶

While we are very focused on the credit implications and their economic impacts as explored above, as investors, we care even more about implications within a portfolio setting. We have covered the correlation trends between major asset classes and categories, and simply put, we put bonds in portfolios to act like bonds. How markets interpret midterm elections and economic data points like Friday’s employment situation report are critical, but bonds are intended to provide protection against deflationary risks (we have to harken back to COVID’s initial outset to recall deflationary risks en masse) while providing income and a correlation benefit offsetting other asset classes, namely public and private equities. How bonds perform amid what will be an action-packed journey into the year-end slipstream will be important, and underscores the importance of portfolio construction and thoughtful planning for clients’ unique situations.

My best to you and your families this holiday weekend.

¹ Merriam-Webster Online Dictionary. “Slipstream Definition.” https://www.merriam-webster.com/dictionary/slipstream. Accessed 3 September 2026. 

² Gonsalves, Antone. “Nvidia Says Supply Shortages are Limiting AI Revenue Growth.” Supply Chain Dive. August 31, 2026. https://www.supplychaindive.com/news/nvidia-says-supply-shortages-are-limiting-ai-revenue-growth/829115/. Accessed 3 September 2026.

³ https://www.hpe.com/us/en/newsroom/press-release/2026/09/hpe-reports-fiscal-2026-third-quarter-results.html

⁴ https://investors.delltechnologies.com/financial-information/quarterly-results

⁵https://investors.broadcom.com/financial-information/quarterly-results

⁶ Gonsalves, Antone. “Nvidia Says Supply Shortages are Limiting AI Revenue Growth.” Supply Chain Dive. August 31, 2026. https://www.supplychaindive.com/news/nvidia-says-supply-shortages-are-limiting-ai-revenue-growth/829115/. Accessed 3 September 2026.

⁷ Millward, Wade Tyler. “Broadcom Q3 Results: CEO Tan Forecasts Surging AI Infrastructure Buildout.” CRN. September 2, 2026. https://www.crn.com/news/ai/2026/broadcom-q3-ceo-tan-forecasts-surging-ai-infrastructure-buildout. Accessed 3 September 2026.

⁸The White House. “G20 Innovation Ministerial Concludes with Consensus Statement.” White House Releases. September 2, 2026. https://www.whitehouse.gov/releases/2026/09/g20-innovation-ministerial-concludes-with-consensus-statement/. Accessed 3 September 2026. ⁹ Bloomberg Intelligence Data, data updated August 19, 2026. Accessed on Terminal 2 September 2026. ¹⁰ Ibid

¹¹ Sha Sajadieh, Loredana Fattorini, Raymond Perrault, Yolanda Gil, Vanessa Parli, Lapo Santarlasci, Juan Pava, Nestor Maslej, Russ Altman, Erik Brynjolfsson, Carla Brodley, Jack Clark, Virginia Dignum, Vipin Kumar, James Landay, Terah Lyons, James Manyika, Juan Carlos Niebles, Yoav Shoham, Elham Tabassi, Russell Wald, Toby Walsh, Dan Weld. “The AI Index 2026 Annual Report,” AI Index Steering Committee, Institute for Human-Centered AI, Stanford University, Stanford, CA, April 2026. https://doi.org/10.48550/arXiv.2606. Accessed 3 September 2026.

¹² Gravison, Jarrod. “Best AI Adoption Statistics 2026: How Many Businesses Actually Use AI?” For AI Things. August 21, 2026. https://foraithings.com/articles/ai-adoption-statistics-2026/. Accessed 3 September 2026.

¹³ International Labour Organization. “Employment and Social Trends 2026.” World of Work Series. https://www.ilo.org/publications/flagship-reports/employment-and-social-trends-2026. Accessed 3 September 2026.

¹⁴ The Federal Reserve Bank of St. Louis. “Delinquency Rate on Credit Card Loans, All Commercial Banks.” August 25, 2026. https://fred.stlouisfed.org/series/DRCCLACBS. Accessed 3 September 2026.

¹⁵ The Board of Governors of the Federal Reserve. “Chairman’s Task Forces for Advancing Monetary Policy.” Federal Reserve Monetary Policy. https://www.federalreserve.gov/monetarypolicy/task-forces.htm. Accessed 3 September 2026.

¹⁶ Northern Trust Wealth Management Research. Data accessed on Bloomberg terminal as of September 3, 2026.