Weekly Investment Update (09/26/2025)

Stocks faced a modest headwind this week as a string of stronger-than-expected economic data challenged the Fed-easing narrative. The biggest highlight came from the labor market, where initial jobless claims fell sharply to 218K, down from 232K and 264K in the prior two weeks — supporting our view of broad labor market stability. In addition, Q2 GDP was revised up significantly to +3.8% (from +3.3%), driven by stronger consumer spending, which jumped from +1.6% to +2.5%. While these prints reinforce the resilience of the economy, they are likely to be seen as hawkish in the near term, creating some pressure for stocks, especially as at least some of the rally has been driven by Fed easing expectations.

Importantly, we do not subscribe to the view that “bad news is good news” simply because it paves the way for more rate cuts. While we do expect further cuts, we believe a stable economic backdrop is a more important foundation for equity markets. 

With stocks at all-time highs, it’s also worth remembering that historical returns are no worse — and often better — when starting from new highs, provided the fundamentals remain strong. In addition to labor market stability, the AI investment cycle continues to accelerate. Micron’s earnings this week reinforced that trend, with the company announcing that AI-related capex will increase in 2026 beyond the $13.5 billion it’s spending this year, further supporting the view that this secular trend remains firmly intact.

Historical Data Shows All-time Highs Are Not Necessarily a Bad Time to Invest

What is happening: On Monday, the S&P 500, Dow Jones, and NASDAQ indexes all closed at new all-time highs. These levels reflect persistent strength in large-cap leadership, particularly among companies tied to AI and cloud infrastructure, such as Nvidia, Oracle, and Broadcom. The recent surge has also been broad-based: Over the past two weeks, the communication services, consumer discretionary, financials, information technology, industrials, and utilities sectors all traded within 1.2% of their respective peaks.

Why it matters: While many stock indexes are at or near all-time highs, it’s worth noting that many fundamentals are as well. The S&P 500’s earnings per share (EPS) are projected to grow more than 10% in 2025 and hit record levels, supported by resilient consumer demand, technology-driven profit growth, and strong corporate balance sheets. Given that 2024 EPS were also a record, any positive earnings growth in 2025 would push earnings to a new high even if the final rate settles below 10%. Operating margins for the index are also expected to reach a new peak above 18% in the next 12 months, supported by cost efficiencies, pricing power, and productivity gains.

For investors, although there can be hesitation around “buying at the top,” historical data suggests this is not necessarily a bad move. Empirical data of S&P 500 returns when investing on all-time-high days show that the return profile is on par if not better than those from random entry points. Over the past 50 years, investing in the S&P 500 on an all-time high day has generated an average 12-month price return of 10.2%, compared to 10.0% when investing on a non-all-time high day.

All-time highs are also a normal part of healthy markets and are not particularly rare in the U.S. equity markets. Over the past 50 years, the S&P 500 index has spent more than 7% of its trading days at new all-time highs — roughly one in 13 trading days. Also, many bull markets tend to accelerate after new all-time highs are reached.

Looking ahead, we believe several macroeconomic forces continue to support equity markets: the Federal Reserve’s shift toward rate cuts, positive real wage growth bolstering consumer demand, and accelerating global investment in technology and AI infrastructure. Together, these dynamics suggest that recent record highs may not represent a near-term ceiling, but rather a potential launching point.

Artificial Intelligence Infrastructure Spending Continues at a Rapid Pace

What is happening: OpenAI and Nvidia this week unveiled one of the most ambitious artificial intelligence (AI) infrastructure initiatives to date, announcing plans to develop more than 10 gigawatts of Nvidia-powered data centers to train OpenAI’s next wave of models such as GPT-5.5 or GPT-6. Once completed, the buildout will require power equivalent to two Hoover Dams, or 4 million U.S. homes. To finance this project, Nvidia has pledged to invest up to $100 billion in OpenAI, with the first tranche expected in the second half of 2026.

The announcement coincides with a recent surge in AI-related investment by the world’s largest technology companies. Oracle and Broadcom recently reported substantial increases in future AI-related revenues, while the “Magnificent Seven” continue to ramp up their capital expenditure on AI. Microsoft, OpenAI’s primary backer, has committed to spending $30 billion this quarter alone to expand its cloud AI capacity. Apple, long seen as lagging in generative AI, recently confirmed it is in discussions with Google to integrate its Gemini model into a revamped Siri assistant, suggesting a potential shift toward outsourcing core AI functions. Together, these moves underscore that success in AI depends as much on securing vast computing resources as it does on software innovation

Why it matters: The scale of Nvidia’s commitment underscores how capital-intensive the AI race has become, with large technology companies viewing the risk of overspending as preferable to the risk of falling behind. Hardware providers, particularly Nvidia, with its dominance in AI chips, are positioned to benefit as demand for computing power continues to climb. Meanwhile, hyperscalers such as Microsoft, Amazon, and Google are financing much of the infrastructure buildout, reinforcing their competitive moats but also raising questions about long-term returns on investment if supply outpaces monetization.

As AI infrastructure investment accelerates, semiconductor, cloud platform, infrastructure, and energy companies will see continued revenue tailwinds. At the same time, valuations may come under scrutiny if expectations for AI monetization lag behind infrastructure outlays, potentially increasing volatility for the S&P 500, where technology-related stocks now account for close to half of the index. Bessemer portfolio managers seek to participate in the AI theme selectively, while maintaining valuation discipline. A recent example is our addition to Alphabet earlier this year, when it was trading at a discount to the broader market.