The recent midterm elections dampened investor optimism around the pace of AI adoption.
Over the past month, many AI-first companies have seen market cap declines of 30% or more, a pullback similar to what we saw in early 2025 when Chinese model provider DeepSeek rattled markets and raised questions about whether the AI capital expenditure paradigm would ever bear fruit. Ultimately, DeepSeek proved to be a head fake, and AI continued to march forward.
This time, however, the dynamics are different. The reality is that political uncertainty and potential moratoriums on data centers will likely slow the pace of AI infrastructure growth, particularly in 2027 and 2028. We believe this shift has been more than priced into the market over the last few trading days, as the broader AI trade has declined by about 20% over that period.
While the near-term trajectory of AI will likely be negatively impacted by the political environment, our long-term outlook remains unchanged: The world will be transformed by AI, and we are still in the early stages of adoption. By ‘early,’ we mean less than 1% adoption across the world’s digital economy and substantially less than 1% adoption in physical AI. With that much runway still ahead, any data centers delayed by the current political environment will ultimately still need to be built to support the remaining adoption.
Personal AI Agents
At Deepwater, we are not just observers of AI but practitioners as well, through our Intelligent Alpha practice and as early adopters of the technology internally. Over the past month and a half, AI agents have dramatically improved our workflows, essentially giving each of us a personal assistant capable of taking away many of the mundane tasks that should ultimately be automated. We believe that personal AI agents remain in the earliest stages of adoption, with an estimated 20 million daily users today, compared with approximately 3.7 billion daily users across Meta’s platforms and more than 3 billion daily Google searches.
Looking back at the recent earnings season, it has become clear from the commentary of Mark Zuckerberg, Elon Musk and Sundar Pichai that personal AI agents will have a dramatic impact on how people interact with technology. We believe these products will evolve from tens of millions of users today to billions of users over time.
Enterprise AI
Separately, our optimism around enterprise AI remains high. Today, average enterprise AI spending is approximately $25 per employee per month, compared with nearly $700 per employee per month in average monthly enterprise software spending. Over time, we believe AI spending will replace and expand upon traditional software spending, creating another significant layer of growth for the industry.
Physical AI
Finally, physical AI has barely begun. Autonomous vehicles and robotaxis have seen ~100% year-over-year growth, yet adoption remains below 1%. Physical AI will extend well beyond autonomous vehicles to autonomous robotic systems across industries ranging from healthcare to logistics. We believe this represents another enormous opportunity that remains largely untapped.
When we put all of this together, we see two realities of AI:
- The near term has become more uncertain, and the slope of AI improvement and infrastructure investment will likely flatten to some degree.
- The long-term opportunity remains enormous. Data centers will need to be built not only to support demand across these emerging use cases but also to maintain America’s geopolitical standing, particularly relative to China. Ultimately, we believe it is in America’s best interest to promote data center development and AI adoption. The invisible hand of the market, combined with the strong hand of the government, should ultimately pave the way for a return to an aggressive ramp in infrastructure and application development in the years ahead.
One final thought on the recent market pullback: We believe this represents a buying opportunity in the broader AI trade. That does not mean every company will be successful, nor do we believe all AI-related companies will be winners. However, we think the delays in data center construction and the potential impact on the pace of AI growth have been more than priced into the market. In our view, this dislocation presents a buying opportunity that we are taking advantage of rather than a change in the long-term story.