Quick View: An argument for the long view during AI volatility

On November 5, the tech-heavy NASDAQ Composite Index registered its worst day since April 2025 as investors continue to parse the significance of the U.S. midterm elections. The catalyst for the sell-off, in our view, isn’t found in divided government returning to Washington but in the states where governorships – and in some cases legislative chambers – switched from Republican to Democratic control.

While affordability and the war with Iran were key issues on the federal level, within states, many of the victorious Democrat candidates incorporated rhetoric that questioned the environmental and societal costs of the ongoing AI data center buildout.

According to the Wall Street Journal, Texas, Georgia, Michigan, and Ohio will all soon have occupants of their governors’ mansions who campaigned against data centers. This matters, as much of the regulatory, permitting, and zoning framework governing these massive infrastructure projects occurs at the state and local level.

Even prior to the 2024 presidential election, we closely monitored the degree to which regulatory risk could impact the AI infrastructure buildout. That threat has grown over the past two years as NIMBYism¹ has taken root in many communities. All along, our research team has reviewed developments on the local level and on a company-by-company basis and assigned an appropriate risk premium to reflect the threat of a more strenuous regulatory framework.

The November 3 result, in our view, creates a thicker veil of policy uncertainty across the entire AI infrastructure value chain. Importantly, however, we believe the AI buildout will continue, but the locations in which it will be concentrated, and the pace of the rollout, are now in question.

An existential matter?

As we anticipated, the midterm elections returned divided government to Washington. The Trump administration remains in control of the executive branch, and from both a geopolitical and economic perspective, it views maintaining AI supremacy among the highest of national priorities. Similarly, the hyperscalers funding this buildout view the continued advancement and integration of AI as fundamental to their businesses.

We believe the gigawatts of AI capacity slated to come online in 2027 will largely continue according to plan. Longer term, projects that run into new regulatory hurdles may end up shopping for friendlier jurisdictions. One could realistically surmise, however, that even in states with a certain degree of data-center hostility, concessions and compromises will be made that address environmental, water usage, and power concerns without having the location give up potential tax revenue. Indeed, since the emergence of NIMBYism, data center companies have dedicated significant resources to community engagement on the state and local level with respect to those concerns.

Reshuffling the landscape

Within technology, a higher level of regulation tends to favor the incumbents. Large companies typically have the financial firepower and staff to withstand and comply with more rigorous oversight. We believe this would be the case should the data center buildout face such headwinds. Hyperscalers, with their existing torrent of cash flows from software sales, cloud services, and digital advertising, would likely be advantaged over AI-centric neocloud providers.

This week’s broad-based, AI-ecosystem sell-off – which hit semiconductor stocks particularly hard – along with the specter of more regulation, presents investors with the opportunity to seek out those business models that have multiple business lines. In our view, these companies are likely to be better positioned compared to single-product companies and those that have only seen an earnings lift through pricing power during this period of acute supply/demand imbalances.

Also potentially disadvantaged are smaller and newer companies and those that are less efficient on a token-per-watt basis. Should data center backlash manifest itself in more restrictive policy, spread to other states, or take root at the federal level  (possibly after the 2028 election cycle), we could expect to see compute capacity shift to other regions. This includes countries that, thus far, have little to no exposure to the AI buildout but would likely welcome it for economic and geopolitical reasons.

A clear destination but a winding path

In closing, as we discussed prior to Tuesday’s election, the AI buildout will unfold across multiple election cycles. The path, not only from a policy perspective but also technological and funding, will invariably be uneven. Still, with the transformative potential of AI becoming more evident every day, especially with respect to productivity gains and the advancement of knowledge, we believe there is no turning back from what we consider the most transformative technology advancement in generations.

A shift in regulatory regime will impact winners and losers. But by a much larger degree, so will the willingness of companies to push innovation and alter their business models to incorporate AI. All these variables are in play, and this election is, in our view, proof that the landscape and ultimate winners can change quickly.

The views presented are as of the date published. They are for information purposes only and should not be used or construed as investment, legal or tax advice or as an offer to sell, a solicitation of an offer to buy, or a recommendation to buy, sell or hold any security, investment strategy or market sector. Nothing in this material shall be deemed to be a direct or indirect provision of investment management services specific to any client requirements. Opinions and examples are meant as an illustration of broader themes, are not an indication of trading intent, are subject to change and may not reflect the views of others in the organization. It is not intended to indicate or imply that any illustration/example mentioned is now or was ever held in any portfolio. No forecasts can be guaranteed and there is no guarantee that the information supplied is complete or timely, nor are there any warranties with regard to the results obtained from its use. Janus Henderson Investors is the source of data unless otherwise indicated, and has reasonable belief to rely on information and data sourced from third parties. Past performance does not predict future returns. Investing involves risk, including the possible loss of principal and fluctuation of value.

Not all products or services are available in all jurisdictions. This material or information contained in it may be restricted by law, may not be reproduced or referred to without express written permission or used in any jurisdiction or circumstance in which its use would be unlawful. Janus Henderson is not responsible for any unlawful distribution of this material to any third parties, in whole or in part. The contents of this material have not been approved or endorsed by any regulatory agency.