Does the Democratic Party oppose AI?

Author: The Beauty of Bayesian

When political slogans meet industrial interests: Through the struggle between the two parties in the United States, we can see the true logic of the AI capital market. After the 2026 midterm elections in the United States, a familiar narrative emerged in the market: the Democrats won, and the American AI industry would suffer; the construction of data centers might be restricted, capital expenditures of tech giants would be impacted, and thus the AI bull market faced systemic risks. This statement seems logically complete. However, if we extend the historical perspective slightly, we will find a very worthy question to ask:

From 2022 to 2024, the President of the United States was the Democrat Joe Biden. What happened to the American AI industry during 2023 to 2024?

It was the explosion of generative AI, a surge in computing power demand, the expansion of capital expenditures by tech giants, and historic profit growth for leading AI companies like Nvidia.

Did these things happen in another America? If the Democrats are inherently opposed to AI development, then why did the most important commercialization explosion period of the American AI revolution occur precisely during the presidency of a Democrat?

Of course, a market rise does not necessarily mean that government policy is correct. However, historical facts at least remind us that directly equating Democratic governance with the decline of the AI industry is logically too simplistic and lacks sufficient empirical support. More important than party labels is the understanding of real policies, economic interests, and industrial transmission mechanisms.

I. First, Respect History: How Much Did AI Actually Rise During the Democratic Administration?

Let’s look at a few sets of numbers.

In 2022, the Nasdaq Composite Index fell by 33.10%.

In 2023, the Nasdaq rose by 43.42%.

In 2024, the Nasdaq continued to rise by 28.64%.

In just the two years of 2023 to 2024, the cumulative increase of the Nasdaq reached approximately 84.5%.

Now let’s look at the most representative company in the AI industry—Nvidia.

According to Nvidia's annual report submitted to the SEC, its data center business revenue:

FY2023: $15.005 billion;

FY2024: $47.525 billion;

FY2025: $115.186 billion.

In just two fiscal years, the revenue from the data center business grew by about 7.7 times. Here, FY2025 ends on January 26, 2025, mainly reflecting the business development of 2024. Meanwhile, Meta's capital expenditure in 2024 (including principal payments for financing leases) reached $39.23 billion, and Microsoft's cash expenditure on property and equipment in FY2024 reached $44.48 billion. These expenditures are not all attributable to AI, but they fully illustrate that large tech companies were making massive infrastructure investments at that time.

These numbers are not speculative but are formally disclosed financial data from the companies. It is important to emphasize that during 2023 to 2024, the United States was under a Democratic president and a politically divided Congress, so the rise cannot be entirely attributed to the Democrats. Similarly, in 2022, when the Democrats controlled the White House and both houses of Congress, tech stocks still fell sharply.

What does this indicate?

The rise and fall of tech stocks have never been a simple function of party variables. The tech bear market of 2022 was closely related to the Federal Reserve's rapid interest rate hikes, rising real interest rates, and valuation compression; the AI bull market of 2023 to 2024 was driven by technological breakthroughs, upward revisions in corporate earnings, expansion of capital expenditures, and a recovery in risk appetite.

If one party could explain the rise and fall of tech stocks, then the bear market of 2022 and the bull market of 2023 under the same president could not be simultaneously explained.

II. Does the Biden Administration Really Oppose AI? Let's See What It Has Actually Done

More worthy of study than market gains are the actual policies of the time.

First, in August 2022, the CHIPS and Science Act was officially signed.

This act allocated $52.7 billion for the U.S. semiconductor industry and related programs, with about $39 billion for semiconductor manufacturing incentives and about $11 billion for research and development. This act received bipartisan support and was not completed solely by the Democrats. However, it was indeed signed and implemented during Biden's administration. One of its core goals is to strengthen U.S. semiconductor manufacturing capacity and reduce the vulnerability of critical technology supply chains.

Semiconductors are an important component of AI computing infrastructure. If we say that the federal government at that time was completely opposed to the development of the AI industry, it would be difficult to explain the existence of such large-scale industrial policies.

Second, in October 2023, Biden signed AI Executive Order EO 14110.

Many people today mention this executive order and only remember safety testing, algorithm risks, and government regulation. However, the complete document not only involves risk management but also includes support for AI innovation, promoting competition, attracting talent, advancing research, and government applications. This executive order reflects a policy combination: it hopes for AI development while also wishing for certain governance over AI. There can be debates about whether the compliance costs are too high or whether some safety requirements stifle innovation. However, we cannot directly rewrite "regulating AI" as "opposing the development of AI technology."

Third, in September 2024, the White House proactively coordinated the construction of AI data centers.

On September 12, the Biden administration convened major cloud providers, AI companies, data center operators, and power companies to discuss large AI data centers, grid construction, clean energy, project approvals, and labor demands. Subsequently, the White House promoted the establishment of an AI Data Center Infrastructure Working Group to coordinate relevant departments to resolve construction bottlenecks. This precisely proves that the Democratic government at that time had realized that AI competition is not only about algorithms but also about GPU, energy, electricity, data centers, and infrastructure delivery capabilities.

Fourth, in October 2024, AI officially entered the core of the U.S. National Security Strategy.

The Biden administration issued NSM-25 National Security Memorandum, requiring the U.S. to maintain its leading position in advanced AI and coordinate computing infrastructure, research and development, and national security applications. Even on January 14, 2025, just before Biden left office, he signed EO 14141, specifically promoting the construction of U.S. AI infrastructure, which included requiring the Department of Defense and the Department of Energy to identify federal lands suitable for building advanced AI data centers.

Of course, executive orders do not mean that projects have already been completed, and policy goals do not equate to final effects. However, at least from the policy texts and administrative actions, the so-called "Democrats oppose the development of AI in the U.S." does not align with these facts.

III. The Real Distinction: Supporting AI Technology Does Not Mean Unconditional Support for Tech Giants

Here, an important distinction needs to be introduced: supporting the development of an industry is not the same as supporting all companies in that industry to retain all profits.

The Biden administration supports semiconductor and AI infrastructure on one hand while also strengthening antitrust measures, AI safety governance, and export controls on the other.

In January 2024, the U.S. FTC officially investigated the partnerships between Microsoft and OpenAI, Amazon and Anthropic, Google and Anthropic, focusing on whether large tech companies were limiting competition through funding, computing power, and contractual advantages.

In 2022 and 2023, the Biden administration also gradually strengthened export controls on advanced computing chips to China. These policies may indeed affect the revenues, costs, and competitive advantages of tech companies and should not be taken lightly. However, they target market structure, national security, and technology diffusion, which does not mean that the U.S. government wishes for its AI industry to stop innovating.

From an economic perspective, there are three different interests:

The nation wants to possess advanced AI technology.

Companies want to create profits through AI technology.

The public wants the costs and benefits brought by AI development to be reasonably distributed.

The three can generally align but cannot always be completely aligned.

Therefore, a government can very well subsidize domestic semiconductor manufacturing while restricting the export of certain high-end GPUs; promote AI data center construction while requiring large tech companies to bear more grid costs. Compressing these different levels of policies into a single statement of "supporting AI" or "opposing AI" would obscure the real issues worth studying for investors.

IV. The Controversy in 2026 Has Essentially Shifted: Who Will Foot the Bill for AI Expansion?

Why has data centers suddenly become an important issue in local elections in the U.S. in 2026?

The answer lies first in the fact that the expansion of AI infrastructure has begun to generate increasingly significant external costs. A study released by the U.S. Department of Energy in 2024 showed that in 2023, data centers across the U.S. consumed approximately 176 terawatt-hours of electricity, accounting for 4.4% of the nation's total electricity consumption. The study predicts that by 2028, data center electricity consumption could reach 325 to 580 terawatt-hours, accounting for 6.7% to 12.0% of the nation's electricity consumption.

Note: This is a scenario forecast at that time, not an already realized electricity consumption, but it clearly reveals the scale of the problem. Data centers not only require GPUs but also need additional power generation capacity, transmission lines, substations, cooling systems, water resources, and land.

Who will bear these infrastructure costs? If large tech companies gain computing power and commercial profits while some of the grid construction costs are passed on to ordinary households through utility rates, local political resistance will naturally rise. This resistance may not necessarily stem from a dislike of AI technology; it could also arise from concerns about electricity prices, land use, environmental protection, and public subsidies.

In January 2026, Democratic Senator Cory Booker and others proposed the "Power for the People Act," requiring large data centers to bear the related costs of grid expansion to reduce the burden on ordinary residents.

However, it is worth noting that this position is not unique to the Democrats. In September 2026, the House of Representatives passed the Republican-led "Ratepayer Protection Act" with 417 votes in favor and 3 against, with many Democratic lawmakers also voting in favor. The bill subsequently failed to reach the 60-vote threshold needed to advance in the Senate with a procedural vote of 57 to 43. Opposing Democratic senators argued that the bill merely required states to consider relevant standards and lacked mandatory enforcement; Republicans emphasized the need to respect state regulatory authority.

This incident illustrates that both sides have a considerable overlap in their goals to protect residents who bear electricity costs, but there are still differences in the strength of constraints and enforcement methods. Real political divisions are often much more complex than the bipartisan opposition summarized by the media.

V. It Cannot Be Denied: The Democratic Candidate's Proposal to Restrict Data Centers May Have Real Impacts

Rational analysis should not swing to the other extreme—believing that all statements restricting AI infrastructure are merely campaign slogans.

After the midterm elections on November 3, 2026, Reuters reported that the winning Democratic candidate for Texas governor, Gina Hinojosa, promised to suspend approvals for new data center projects until clearer community and consumer protection rules were established. Texas holds a significant share of the planned data center capacity in the U.S. Therefore, even if the approval suspension does not affect the final AI demand, it may change project delivery timelines, regional allocations, and financing returns.

However, another fact is equally noteworthy: before this election, Republican Texas Governor Greg Abbott had already taken measures to suspend some new grid access and strengthen scrutiny of data center projects. In September 2026, at least five Republican candidates in various states proposed or strengthened their positions on data center restrictions. This makes it insufficient to fully explain local data center controversies as "Democrats opposing AI."

Whether the new governor's campaign promises are fulfilled also depends on administrative authority, state legislatures, public utility commissions, existing contracts, and judicial reviews. Media announcements of victory do not equate to the related restrictions being formally enacted. Opposing the siting or subsidies of certain data center projects is not the same as opposing AI itself. However, it cannot be ignored that even if long-term AI demand remains strong, project delays may cause real cash flow losses. For highly leveraged data center developers reliant on project operation and ongoing financing, this risk is particularly significant.

VI. The Common Economic Constraints of the Two Parties in the U.S. on AI Issues Cannot Be Ignored

AI is no longer just a software technology issue.

It relates to semiconductor manufacturing, national security, military applications, cloud computing, energy investment, scientific research competition, and the competitive position of large American tech companies in the global market. This means that whichever party holds political power must face multiple simultaneous economic incentives.

On one hand, advanced AI may enhance productivity, expand economic output, and create new industrial opportunities. On the other hand, AI infrastructure requires substantial electricity, capital, and public resources, and may also bring employment shocks, market concentration, and cost-shifting issues.

The two parties have different levels of emphasis and policy tools regarding these issues, but it is impossible to eliminate these constraints with just a campaign slogan.

There is also room to discuss the personal investment holdings of political figures. The Pelosi family, some Democratic lawmakers, and relevant congressional members have observable tech stock exposure through trading-tracking ETFs, which is a line of interest worth studying. However, public holdings face issues such as disclosure delays, wide ranges of amounts, and the need to distinguish between family members' and lawmakers' own holdings. One cannot simply assert that a lawmaker will not support regulations affecting data center profit margins just because their family holds NVDA.

Investment holdings can provide information on economic incentives but cannot replace direct analysis of legislative voting, executive actions, and policy implementation. More important than observing what political figures have in their pockets is examining what policies they ultimately enact.

VII. The Most Common Mistake in Capital Markets: Equating Policy Narratives Directly with Profit Changes

From an investment perspective, negative political news truly affects stock value only after passing through a transmission chain: political events → policy texts → executive implementation → project economics → corporate profit expectations → valuations and risk premiums.

Any link in this chain can be interrupted or may differ from the market's initial speculation. For example, if a state increases the grid access costs for data centers, it may lead to a decline in developers' project IRR; however, if the cloud vendors ultimately bear more energy costs, GPU demand may not necessarily decline in tandem.

Similarly, project delays may negatively impact specific regional data center developers but make existing facilities with sufficient power and access capacity even scarcer. If the capital market simplifies all these differences to "negative for the AI sector," it may lead to mispricing among different companies.

Using a causal attribution framework, risks can be broken down into three competing hypotheses:

H1: The final demand for the AI industry is substantially suppressed by policy. Evidence is needed to observe companies cutting capital expenditures, canceling orders, and revising down computing power demand.

H2: AI demand still exists, but the costs of infrastructure construction are being redistributed. Evidence is needed to observe grid access fees, project delays, developer profit margins, and investment returns.

H3: Political events are merely triggers for market repricing. What truly amplifies the decline may be overvaluation, crowded positions, deleveraging, options hedging, and insufficient liquidity.

The third scenario is particularly worth being cautious about.

The experiences from 2022 to 2024 have already shown that corporate profits, monetary policy, and valuation changes can produce vastly different market outcomes within the same political environment.

The AI market in 2026 will be similar. Even if long-term technological demand does not change, substantial capital expenditures, financing costs, and high beta positions can trigger severe adjustments.

Therefore, being long-term bullish on AI does not mean that every drop should be immediately bought, nor does it mean that portfolio risks can be ignored.

A truly valuable "Buy the dip" should be based on independent judgments of fundamentals, valuations, and liquidity, rather than simple optimism or pessimism towards a particular political party.

VIII. The Most Worthy Reflection Is Not the Democratic Party's Attitude Towards AI, but the Analysis Method of Investors

Looking back at the entire AI cycle, we find a common cognitive issue:

The market likes to use simple political labels to explain complex industrial changes.

If the Democrats win, AI may be regulated, so tech stocks will fall.

If the Republicans win, regulation may be relaxed, so tech stocks will rise.

This narrative spreads efficiently but often overlooks the most important intermediate variables in asset pricing.

The AI bull market from 2023 to 2024 did not stop due to the Biden administration's implementation of AI safety regulations; the AI infrastructure controversies in 2026 will not automatically disappear just because the Republicans support American technological competitiveness.

Policies may change construction costs but do not necessarily change demand; they may harm the profits of certain companies but not the entire industry; they may also lead to real project delays that cannot simply be labeled as market noise.

Truly rigorous investment research must model party positions, policy constraints, project economics, corporate profits, and stock valuations separately.

Looking back at the initial question: Does the Democratic Party oppose AI?

At least based on the federal policy records from 2022 to 2024, one cannot conclude that the Democratic Party as a whole opposes the development of the AI industry. However, this does not mean that there are no political forces within the Democratic Party advocating for strict limitations on certain AI applications and data center construction. A more accurate understanding is that the U.S. AI industry has entered a new phase where technological expansion and social cost distribution are intertwined.

For investors, what determines long-term returns is not labeling a party as "supporting AI" or "opposing AI," but rather whether technological demand can continue to grow, whether companies can generate free cash flow, and which parts of value distribution policies have actually changed.

Do not treat political slogans as industrial facts, do not treat election results as profit forecasts, and do not treat market sentiment as investment conclusions.

If the history from 2022 to 2024 leaves any important insight, it is this:

The long-term development of the AI industry can transcend different political cycles, but the valuations and profits of AI companies must still undergo the reality check of each political cycle.

True independent thinking is not believing that any political party is always favorable to tech stocks, but rather being able to identify industry trends, verify policy evidence, and judge where capital will flow, regardless of who is in power.