Market Update—Quarter Ending December 31, 2025

The information technology sector delivered an outsized gain through the first 11 months of 2025 in a broadly “risk-on” environment, except for the brief stretch of “tariff turmoil” in March and April.

For 2026, I see continued strong spending on AI infrastructure by hyperscalers and enterprises. Semiconductors should play a starring role in this scenario, as AI models evolve with enhanced capabilities, giving them more powerful ways to parse huge datasets. The “picks and shovels” that have brought the AI train this far—graphics processing units, high-speed memory and data centers—will continue to be integral to successive improvements in 2026 and beyond. The flip side of this massive AI spending is that other segments of technology, such as IT services, will likely see reduced capital expenditures, given that they are competing with AI for dollars in limited corporate IT budgets. Elsewhere, software companies could be at risk of major disruption, if AI models increasingly incorporate application capabilities.

2025: Information technology led the way

After a strong year in 2024, tech stocks began 2025 on the back foot. The broader stock market saw post-election optimism give way to investor concern about the impact of a flurry of executive actions by the new presidential administration, especially related to tariffs. Within technology specifically, relatively high valuations to start the year in AI-related stocks and uncertainty caused by DeepSeek’s announcement near the end of January were headwinds for the group. DeepSeek, a Chinese generative artificial intelligence startup, released a free large language model that was cheaper and faster to develop than existing models, raising questions about spending on AI infrastructure and the valuations of the biggest players, including chipmakers.

The administration’s rollout of wide-ranging “reciprocal tariffs” in early April prompted a sharp drop in the broad-market indexes at that time. Just as quickly, though, the administration subsequently paused most tariffs for 90 days, setting the stage for a strong market rebound that carried well into the fourth quarter of 2025. Tech stocks recovered their mojo during this advance, resuming market leadership, especially as hyperscalers reaffirmed and increased their spending commitments on AI.
​

Year-to-date price return

2026: AI models keep evolving and spending continues apace

At this point, I see no prospect of flagging AI spending, given the importance of the technology across the market. Because of this, I expect many of the AI-exposed stocks that have led the market over the past year to continue to outperform in the year ahead, especially chip companies with AI exposure. I don’t agree with the perception that the “picks and shovels” phase of AI will go away at some point, when AI applications become more widespread and polished. Spending on picks and shovels should be with us indefinitely, as AI models constantly evolve and require updated equipment to support new capabilities. The more software-dependent DeepSeek model referenced earlier has been widely available now for almost a year, and while it can compete with ChatGPT and other hardware-intensive AI platforms in some respects, it has not materially taken market share from them, and I don’t expect it to do so in 2026. This is good news for AI infrastructure heavyweights such as Nvidia and Taiwan Semiconductor Manufacturing.

Nvidia has been highlighted so often in connection with AI that some investors might be tiring of hearing about it or wonder if some other company is set to swoop in and steal its thunder. In my view, though, Nvidia merits all the attention it gets because it really is that far ahead of the competition. No other company is close to offering products with the combination of integrated hardware, software and algorithms that Nvidia does, especially since its Blackwell chipset was introduced in 2024. Blackwell was developed specifically for use in high-performance graphics processors and delivers significant improvements in performance and energy efficiency. Nvidia’s Vera Rubin superchip, the company’s next generation of AI hardware, should be in production sometime in 2026.

Meanwhile, Taiwan Semiconductor Manufacturing remains the world’s leading foundry for cutting-edge chips suitable for AI applications. With its technological superiority and client list including the likes of Apple and Nvidia, it’s no wonder that, as of the third quarter of 2024, the company held a roughly 65% share of the global semiconductor foundry market.

There is one non-AI chip company I’ll highlight as well: GlobalFoundries. As of late 2025, the Trump administration is considering applying Section 232 tariffs to companies that make certain products with less than 50% of manufacturing done on U.S. soil. The Chinese, having been denied full access to U.S. cutting-edge chip technology, have been aggressively trying to take market share in trailing-edge chips. This new tariff policy is an attempt to blunt that effort and maintain a market for U.S.-made trailing-edge chips. With its foundries in New York and Vermont, GlobalFoundries is positioned to take advantage of this effort to strengthen domestic chip manufacturing, which is considered a strategically important industry by the federal government. The company also has foundries in Germany and Singapore, positioning it to assist other countries in fending off Chinese hegemony in trailing-edge chip production. A separate tailwind for the company could come in the form of a winding down of the inventory glut hampering the market for non-AI chips, a process I believe is near completion after lasting roughly three years.

Prospects for other technology themes and groups in 2026

With all that said, I’m covering the tech sector as a whole and continue to invest in other themes that are impacting the sector, such as cloud computing, digital transformation and digital security. I’m being very selective about investments in software, an industry currently being disrupted by AI. This disruption is largely due to a significant shift over the past year in our understanding of AI’s application layer. This layer is the most user-facing part of the AI stack, embedding AI capabilities into software applications, products and services. Increasingly, the market has come around to the view that newer AI models are going to include an application layer, putting software companies providing applications at the heart of the expected disruption.

I’ll continue focusing on situations where my analysis diverges from the consensus view and look for opportunities within the sector that I believe offer the best relative risk/reward. Right now, though, AI is where the vast majority of corporate capital expenditures are being deployed, and I think it’s more prudent to ride that wave than try to swim against the tide.