Author: The Beauty of Bayesian

When political slogans meet industrial interests: Through the lens of the American two-party struggle, 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; data center construction might be restricted, the capital expenditure of tech giants would be impacted, and thus the AI bull market faced systemic risks. This claim seems logically sound. However, if we stretch the history a bit, we find a very worthwhile question to probe:

From 2022 to 2024, who was the President of the United States? It was the Democrat Biden. What did the American AI industry experience from 2023 to 2024?

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

Did these events take place in another America? If the Democrats are inherently opposed to AI development, then why did the most significant commercialization explosion period of the American AI revolution happen precisely during the tenure of a Democrat president?

Of course, a market increase does not equal that government policy is necessarily correct. However, historical facts at least remind us: it is too crude to directly equate Democratic rule with the decline of the AI industry, and there is insufficient empirical support for such a conclusion. More important than party labels is an understanding of real policies, economic interests, and industrial transmission mechanisms.

1. First Respect History: How Much Did AI Actually Increase During the Democratic Presidency?

Let’s first 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 and 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 was:

FY2023: $15.005 billion;

FY2024: $47.525 billion;

FY2025: $115.186 billion.

In just two fiscal years, the revenue of the data center business increased by about 7.7 times. The FY2025 figure is as of January 26, 2025, primarily reflecting the business development of 2024. Meanwhile, Meta's capital expenditure in 2024 (including principal payments on financing leases) reached $39.23 billion, and Microsoft's cash expenditures on property and equipment in FY2024 reached $44.48 billion. These expenditures are not entirely related to AI, but they clearly indicate that large tech companies were making massive infrastructure investments at the time.

These numbers are not speculative assumptions but official financial disclosures from companies. It is essential to emphasize that between 2023 and 2024, the United States was under a Democratic president with a divided Congress, which prohibits attributing the rise solely to the Democrats. Similarly, in 2022, when the Democrats controlled the White House and both chambers of Congress, tech stocks still fell sharply.

What does this indicate?

Tech stock fluctuations have never been a single-variable function of party affiliation. The tech bear market in 2022 was closely related to the Federal Reserve's rapid interest rate hikes, rising real interest rates, and valuation compression; while the AI bull market in 2023 and 2024 was jointly driven by technological breakthroughs, upward revisions in corporate profits, expansion in capital expenditures, and a recovery in risk appetite.

If a single party could explain tech stock movements, then the bear market of 2022 and the bull market of 2023 under the same president could not be explained simultaneously.

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

What’s more worth studying than market increases are the actual policies of that year.

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

This act allocated $52.7 billion for the American semiconductor industry and related programs, with approximately $39 billion for semiconductor manufacturing incentives and about $11 billion for R&D. This act received bipartisan support and was not solely the achievement of the Democrats. However, it was indeed signed and implemented during Biden’s presidency. One of its core goals is to strengthen U.S. semiconductor manufacturing capacity and reduce vulnerabilities in critical technology supply chains.

Semiconductors are a vital component of the AI computational infrastructure. If one claims that the federal government was entirely opposed to the development of the AI industry at that time, it would be challenging to explain the existence of such large-scale industrial policies.

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

Many people today refer to this executive order and only remember security testing, algorithmic risks, and government regulation. However, the complete document encompasses not only risk management but also supports for AI innovation, promotes competition, attracts talent, and encourages research and governmental applications. This executive order reflects a policy mix: it hopes for AI development while also wishing for some governance over it. One can argue whether its compliance costs are too high or whether certain safety requirements impede innovation, but one cannot directly rewrite "regulating AI" as "opposing AI technology development."

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

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 workforce needs. 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 demonstrates that the Democratic administration of that time recognized: AI competition is not only about algorithm competition, but also about GPU, energy, power, data center, and infrastructure delivery capabilities.

Fourth, in October 2024, AI officially entered the core of the U.S. national security strategy.

The Biden administration released the NSM-25 National Security Memorandum, calling for the United States to maintain its leading position in advanced AI and to coordinate computing infrastructure, R&D, 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 a requirement for the Department of Defense and the Department of Energy to identify federal land suitable for advanced AI data center construction.

Of course, an executive order does not mean projects have been completed, and policy goals do not equate to final outcomes. But at least from the perspective of policy texts and administrative actions, the so-called “Democrats oppose the development of AI in the U.S.” does not align with these facts.

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

Here, it is necessary to introduce an important distinction: 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 strengthening tech antitrust, AI safety governance, and export controls on the other.

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

In 2022 and 2023, the Biden administration also gradually intensified export controls on advanced computing chips to China. These policies may indeed affect tech companies' revenues, costs, and competitive advantages, which cannot be glossed over. But they target market structures, national security, and technological diffusion, which does not equal that the U.S. government desires for its own AI industry to stop innovating.

From an economic perspective, there are three different interests at play:

The state wants advanced AI technology.

Companies want to create profits through AI technology.

The public wants a reasonable distribution of costs and benefits brought by AI development.

These three can generally align, but they cannot always be completely consistent.

Therefore, it is entirely possible for a government to subsidize domestic semiconductor manufacturing while restricting exports of some high-end GPUs, to promote AI data center construction while requiring large tech companies to bear more costs for the grid. Compressing these different levels of policy into a single statement of “support for AI” or “oppose AI” can obscure the real questions that investors should be studying.

4. The Controversy of 2026 has Essentially Shifted: Who Will Foot the Bill for AI Expansion?

Why did data centers suddenly become an important issue in local elections in America in 2026?

The answer lies primarily in the rising external costs generated by the expansion of AI infrastructure. A study released by the U.S. Department of Energy in 2024 showed that in 2023, data centers in the U.S. consumed approximately 176 terawatt-hours of electricity, accounting for 4.4% of the nation's total electricity consumption. The study projected that by 2028, data center electricity consumption could reach between 325 and 580 terawatt-hours, accounting for 6.7% to 12.0% of national electricity consumption.

Note: This is a scenario projection from that time, not an already achieved level of electricity consumption, but it clearly reveals the magnitude of the issue. Data centers not only require GPUs, but also need new power generation capacity, transmission lines, substations, cooling systems, water resources, and land.

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

In January 2026, Democratic Senator Cory Booker and others introduced 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 noteworthy that this position is not unique to the Democrats. In September 2026, the House of Representatives passed the Republican-led “Ratepayer Protection Act” with a vote of 417 in favor to 3 against, and many Democratic representatives also voted in favor. The bill subsequently failed to reach the necessary 60 votes for advancement in the Senate in a procedural vote of 57 to 43. Opposing Democratic senators argued that the bill only required states to consider relevant standards and lacked mandatory enforcement; Republicans emphasized the need to respect state regulatory authority.

This event illustrates that both sides have a considerable overlap in their objective of protecting residents who bear electricity costs, yet there are still disagreements on the stringency of constraints and enforcement methods. The real political divisions are often much more complex than the two-party opposition summarized by the media.

5. It Cannot Be Denied: The Democratic Candidate's Position on Data Center Restrictions Could Have Actual Impact

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

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

However, another fact is equally worth noting: before this election, Republican Texas Governor Greg Abbott had also taken measures to suspend the approval of some new grid connections and intensified scrutiny over data center projects. In September 2026, at least five Republican candidates in various states proposed or intensified their positions on data center restrictions. This makes the explanation of local data center controversies purely as “Democrats against AI” lack a sound basis.

Whether the new governor's campaign commitments will be fulfilled also hinges on executive authority, state legislatures, public utility commissions, existing contracts, and judicial reviews. Media announcements of victories do not equate to the related restrictions officially taking effect. Opposing certain data center project sites or subsidies is not the same issue as opposing AI itself. However, it cannot be overlooked: even if long-term AI demand remains strong, project delays may still result in real cash flow losses. For developers of data centers that rely heavily on leverage, project operation, and ongoing financing, this risk is especially significant.

6. There Exists a Common Economic Constraint That Must Not Be Ignored Between the Two Parties on AI Issues

AI is no longer simply a software technology issue.

It relates to semiconductor manufacturing, national security, military applications, cloud computing, energy investment, research competition, and the competitive positioning of large U.S. tech companies in the global market. This means that regardless of which party holds political power, they must face a variety of 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 vast amounts of electricity, capital, and public resources, and it may also bring challenges such as job impacts, market concentration, and cost shifting.

Both parties place different levels of emphasis on these issues and employ different policy tools, but neither can eliminate these constraints based solely on a campaign slogan.

Here, the personal investment holdings of political figures can also be discussed. The Pelosi family, some Democratic representatives, and relevant congressional members have observable tech stock exposure through trading-tracking ETFs, which is a clue worth researching concerning interests. However, public holdings have issues such as disclosure lags, a wide range of amounts, and the need to distinguish between family members and the representatives themselves. It is also not valid to assert that because a certain representative's family holds NVDA, that they will never support regulations affecting data center profit margins.

Investment holdings can provide information about economic incentives but cannot replace the direct analysis of legislative votes, executive actions, and policy implementations. More important than observing what political figures have in their pockets is verifying what policies they ultimately bring into effect.

7. The Most Common Mistake Capital Markets Make: Equating Policy Narratives Directly with Profit Changes

From an investment perspective, negative political news affects stock value only after passing through a chain of transmission: Political events → Policy texts → Administrative execution → Project economics → Corporate profit expectations → Valuation and risk premiums.

Any link in this chain can break down, or it may differ from the market’s initial speculation. For example, if a state increases the cost of grid access for data centers, it may lead to a decline in developer project IRR; but if in the end, it is merely cloud providers who bear more energy costs, GPU demand may not necessarily decline correspondingly.

Additionally, project delays may negatively impact specific regional data center developers, but they could make existing facilities with adequate power and access capacity even scarcer. If capital markets simplify all these differences to “negative for the AI sector,” there may be mispricing between different companies.

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

H1: The ultimate demand for the AI industry is substantively suppressed by policy. Evidence of companies cutting capital expenditures, canceling orders, and revising down computational demand needs to be observed.

H2: AI demand still exists, but the costs of infrastructure construction are being reallocated. Evidence of grid access fees, project delays, developer profit margins, and investment returns needs to be observed.

H3: Political events are merely trigger points for market repricing. What may truly amplify declines are overvaluation, crowded positions, deleveraging, options hedging, and insufficient liquidity.

The third scenario is particularly worth being cautious about.

Experiences from 2022 to 2024 have demonstrated that corporate profits, monetary policy, and valuation changes can produce drastically different market outcomes under the same political environment.

The AI market in 2026 is likewise. Even if long-term technical demand remains unchanged, huge capital expenditures, financing costs, and high beta positions can trigger severe adjustments.

Thus, a long-term bullish outlook on AI does not mean that every decline should be immediately bought, nor does it mean that overall portfolio risks can be ignored.

Genuine valuable buying the dip should be based on independent judgments of fundamentals, valuations, and liquidity, rather than simple optimism or pessimism towards a political party.

8. The Most Worthy Reflection Is Not the Democratic Party's Attitude Towards AI, but the Investors' Analysis Methods

Looking back over the entire AI cycle, we observe a common cognitive issue:

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

The Democrats won, so AI might be regulated, hence tech stocks will fall.

The Republicans won, so regulation might be relaxed, thus 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 implementing AI safety regulations; nor will the AI infrastructure controversy in 2026 automatically disappear merely because the Republicans support U.S. technological competitiveness.

Policies may alter construction costs but do not necessarily change demand; they may harm the profits of certain companies, but not impact the entire industry; and they may also bring about real project delays, which cannot simply be labeled as market noise.

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

Reflecting on the initial question: Does the Democratic Party oppose AI?

Based on the federal policy records from 2022 to 2024, it cannot be concluded that the Democrats as a whole oppose the development of the AI industry. However, this does not mean that there are no political forces within the Democratic Party advocating strict limitations on certain AI applications and data center constructions. A more accurate understanding is: the U.S. AI industry has entered a new phase where technological expansion intertwines with the distribution of social costs.

For investors, what determines long-term returns is not labeling a political party as “supporting AI” or “opposing AI,” but examining whether technological demand can sustain growth, whether companies can generate free cash flow, and how policies have altered value distribution.

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

If the years 2022 to 2024 imparted any significant revelation, it would be this:

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

True independent thinking is not about believing that any political party will always favor tech stocks, but rather about recognizing industry trends, verifying policy evidence, and determining where capital will ultimately flow, regardless of who is in power.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到[email protected]，本平台相关工作人员将会进行核查。