A Look Back and then Forward

For context, companies within the semiconductor pantheon fell sharply this week, especially overnight on Thursday. South Korean chipmakers SK Hynix and Samsung felt acute selling pressure, with SK Hynix falling 26% in the last five trading days ending October 15 and Samsung shedding 22%. While broad semiconductor indices (including the Philadelphia Semiconductor Index) have not rebounded above late-June peaks, a few catalysts drove this week’s selloff.

First, China’s ChangXin Memory Technologies (CXMT) announced progress on more complex chips occupying less surface area by the end of the year, threatening the above incumbents plus Micron.¹ Second, Taiwan Semiconductor (TSM), the world’s largest semiconductor production fab, delivered strong revenue and earnings in its October 15 release, but profitability declined modestly due to a ramp-up in advanced technology. The company also anticipates slight margin deterioration for the rest of this year versus prior guidance.² On a forward-looking basis, TSM added that its overseas fabrication expansion plans will further impact profitability.

Our working hypothesis remains that as artificial intelligence continues to diffuse across commercial, individual and governmental use cases, more competition will emerge. Several companies have enjoyed years of modest competition, if any, and as this transformative technology continues to evolve, weeks like this will occur. The fits and starts of new entrants, existing providers adjusting their strategy and expenditure plans, and other variables will drive fluctuations within a still-ascending demand environment. We will cover broader forward-looking views later in this piece.

Anthropic is a leading developer of frontier AI models with stated plans for an initial public offering at some point this year, subject to change. The company hosted a presentation at its San Francisco headquarters on Wednesday, highlighting that it achieved profitability on an adjusted basis this quarter and that third-quarter revenue touched $16.8 billion.³ While still a private company, some market observers expressed concerns about the company’s sequential growth rate slowing, with risks that some current clients may opt to develop their own AI models.⁴

CoreWeave is often referred to as a “neo cloud” or tier-2 hyperscaler, providing AI cloud computing power to AI labs, startups and enterprise customers. The company’s equity and debt fell this week, along with data center bonds where CoreWeave is a tenant. With many investors anticipating increased debt issuance to fund AI spend, as we have shared in prior publications, the risk of a higher cost of capital for the AI ecosystem has investor attention. While we do not provide individual security recommendations on these pages, this week represented a shift in risk appetite away from corporate bonds and toward government bonds, as we will detail more in the interest-rate section below. Although credit spreads, or the additional compensation investors demand from corporate issuers relative to government issuers, remain historically compressed, borrowing costs rose for riskier credits this week and, in particular, for those directly exposed to tech infrastructure buildouts.⁵

The bond market remains a central topic for these pages, and this week represented a near-term shift, as noted above. As we’ve shared, we anticipate a tempering of bond yields, noting that the recent sharp moves higher can be attributed to several factors including persistent inflation, concerns about global indebtedness (France being a recent flashpoint), and ample supply of both government and corporate issuance. We noted last week that U.S. government bond auctions of longer-maturity debt went well, with strong demand despite large announced offerings across the corporate landscape. Despite relatively strong domestic consumer and producer price readings on Wednesday and Thursday, core and municipal bonds rallied sharply on Thursday.^8,9

However, as noted earlier, lower-quality issuers and AI-related debt did not enjoy lower interest rates. High-yield credit spreads matched their widest levels since March of this year (levels not seen since post-Liberation Day in May 2025). The riskiest slice of the corporate high-yield index requires more than a 10-percentage-point cushion over Treasury bonds, which is 1.5 percentage points greater than average spreads going back to 1993.¹⁰ Should we see ongoing pressure in the bond market for both AI-related borrowings and traditionally riskier issuer ratings, that may suggest investors are questioning corporate discipline and the supply/demand continuum writ large.

While midterms are important, we think they will likely set up a more consequential 2028 presidential election outcome. Most of the issues on the table concentrate on state and local considerations, especially around data center build-out plans. Should capital markets continue to allow AI ecosystem expansion, public reactions to allowing them are striking and cut across party lines. A September UMass Amherst poll found that 65% of respondents oppose an AI data center in their local community, with only 11% supporting.¹¹ Pew Research came to a similar conclusion and noted that the overall concern about adding data centers cuts across party lines, with 54% of Republicans and 68% of Democrats uncomfortable with adding a new data center near their home.¹² Election outcomes will have an impact on company growth plans, and given AI’s outsized contributions to corporate earnings, state and local outcomes will be key.

Beyond the AI considerations, the election’s implications for future trade policy and indebtedness are key focal areas for us. Global trade continues to shift away from large blocs and toward individual country “deals” and transactions. While the World Trade Organization expects global trade will increase by 4% this year, trade complexion continues to change and inflationary risks have increased.¹³ Changes to Congressional leadership could alter the current administration’s path on trade, but changes to the executive branch in 2028 will also prove pivotal. On indebtedness, we have experienced recent bond market angst across most sovereign borrowers this year and a stark dichotomy between interest rates for highly levered governments versus those with lower debt balances. For the coming election, should we see material shifts in spending policy, the bond market’s response will likely be less than favorable.  

As investors, we seek to build durable portfolios that can handle risk factors associated with changing policies. Recognizing that political agendas on campaign trails can differ from political agendas once leaders assume office, we are most focused on inflationary risks and potential constraints on AI ecosystem build-outs. We will keep you posted on developments but perhaps the best advice we can provide is the timeless “keep your politics out of your portfolio.” Making investment decisions based on perceived policy outcomes from local, state or national elections rarely proves prescient, and sticking with a well-established investment policy or financial plan has our vote.

¹ Downing, Shane. “CXMT president says company’s first 4F² DRAM will be a server DDR5 RDIMM due by year-end — vertical transistors cut each memory cell’s footprint by a third on paper, but no speeds or capacities yet.” Tom’s Hardware. October 13, 2026. https://www.tomshardware.com/pc-components/dram/cxmt-president-says-companys-first-4f2-dram-will-be-a-server-ddr5-rdimm-due-by-year-end-vertical-transistors-cut-each-memory-cells-footprint-by-a-third-on-paper-but-no-speeds-or-capacities-yet. Accessed 16 October 2026.

² Taiwan Semiconductor Manufacturing Company (TSMC). “Edited Transcript: Q3 2026 Taiwan Semiconductor Manufacturing Company Limited Earnings Call.” TSMC Financial Results. October 15, 2026. https://investor.tsmc.com/english/quarterly-results/2026/q3. Accessed 16 October 2026.

³ Lipschultz, Bailey and Natasha Mascarenhas and Rebecca Torrence and Shirin Ghaffary. “Anthropic Revenue Jumps to $16.8 Billion in Quarter Before IPO.” Bloomberg Technology. October 15, 2026. https://www.bloomberg.com/news/articles/2026-10-14/anthropic-revenue-jumps-to-16-8-billion-in-quarter-before-ipo. Accessed 16 October 2026.

⁴ PYMNTS. “Anthropic Quarterly Revenue Jumps 45% Before Planned IPO.” PYMNTS News. October 15, 2026. https://www.pymnts.com/news/investment-tracker/ipo/2026/anthropic-quarterly-revenue-jumps-45-before-planned-ipo/. Accessed 16 October 2026.

⁵ Northern Trust Wealth Management Research, Bloomberg. Data accessed on terminal October 16, 2026.

⁶ Meritshot. “Cisco Case Study- The $2.25 Billion Write-Down That Saved a Company.” February 20, 2026. https://www.meritshot.com/case-studies/cisco. Accessed 16 October 2026.

⁷ Ibid

⁸ U.S. Bureau of Labor Statistics. “Consumer Price Index Summary.” October 14, 2026. https://www.bls.gov/news.release/cpi.nr0.htm. Accessed 16 October 2026.

⁹ U.S. Bureau of Labor Statistics. “Producer Price Index Summary.” October 15, 2026. https://www.bls.gov/news.release/ppi.nr0.htm. Accessed 16 October 2026.

¹⁰ Northern Trust Wealth Management Research, Bloomberg. Data as of 16 October 2026.

¹¹ Brown, Alex. “Data center backlash set to sway midterm elections.” Stateline. October 12, 2026. https://stateline.org/2026/10/12/data-center-backlash-set-to-sway-midterm-elections/. Accessed 16 October 2026.

¹² Kennedy, Brian. “Americans’ views of data centers have turned more negative.” Pew Research Center. September 22, 2026. https://www.pewresearch.org/short-reads/2026/09/22/americans-views-of-data-centers-have-turned-more-negative/. Accessed 16 October 2026.

¹³ World Trade Organization. “Global Trade Outlook and Statistics.” October 2026. https://www.wto.org/english/res_e/booksp_e/gtos1026_e.pdf. Accessed 16 October 2026.