Broadcom delivered exactly what Wall Street typically rewards: strong earnings, robust guidance, and eye-catching growth tied to artificial intelligence. Yet the market response was brutal.

Shares of the chipmaker plunged 11% on Friday, marking their worst single-day drop since January, as investors abruptly pulled back from some of the most crowded names in the AI trade. The selloff spilled across the sector. Oracle fell another 4% a day after tumbling 10% following its own earnings report, while AI-focused infrastructure players such as CoreWeave also came under heavy pressure.

The broader market felt the impact. The Nasdaq slid about 1.4%, while the S&P 500 dropped close to 1%, underscoring how deeply AI-linked stocks have become intertwined with overall market sentiment in 2025.

At the heart of the move is a growing unease that the AI infrastructure boom, which has powered both corporate profits and stock prices for the past two years, may be entering a more complicated phase. Hyperscalers are still pouring money into data centers and custom chips, but investors are increasingly asking how sustainable the pace of spending is, and at what cost to margins and balance sheets.

Broadcom sits squarely in the middle of that debate. The company has been one of the biggest beneficiaries of AI’s rise, supplying custom chips and networking hardware to some of the world’s largest technology firms. Its market capitalization roughly doubled in each of the past two years before extending gains again in 2025, leaving the stock up about 75–80% year to date before Friday’s drop.

“This stock is up 75–80% year to date. You’re seeing a little bit of a pullback,” Vijay Rakesh, an analyst at Mizuho, said on CNBC.

He added that the firm would be buyers on the weakness, raising its price target to $450 from $435. Broadcom was trading below $364 by Friday afternoon.

The numbers themselves were hard to fault. Broadcom reported quarterly revenue of $18.02 billion, beating the $17.49 billion consensus estimate compiled by LSEG. Revenue grew 28% year on year, driven largely by a 74% surge in AI chip sales. Adjusted earnings per share came in at $1.95, comfortably ahead of expectations of $1.86.

Chief executive Hock Tan said the momentum is set to continue. Broadcom expects AI chip sales in the current quarter to double from a year earlier to $8.2 billion, supported by demand for both custom AI accelerators and networking semiconductors used in large-scale data centers.

The company also disclosed a massive $73 billion backlog of AI-related orders over the next 18 months, highlighting the depth of commitments from customers racing to secure compute capacity. That backlog includes $21 billion in orders from Anthropic, which Broadcom named as a key customer.

Still, investors zeroed in on the fine print. One major concern is margin pressure, at least in the near term. Chief financial officer Kirsten Spears warned on the earnings call that “gross margins will be lower” for certain AI chip systems because Broadcom must purchase more components upfront to build complete server racks. In a market where expectations are already sky-high, even temporary margin compression can be enough to trigger a selloff.

There was also disappointment around OpenAI. While Broadcom has touted a multibillion-dollar agreement announced in October, Tan tempered expectations, telling investors that “we do not expect much in ’26,” cooling hopes that the deal would materially boost revenue in the near term.

Bernstein analyst Stacy Rasgon described the market reaction as driven by “AI angst” rather than fundamentals.

“Frankly we aren’t sure what else one could desire as the company’s AI story continues to not only overdeliver but is doing it at an accelerating rate,” Rasgon wrote, reiterating a buy rating and lifting his price target.

The skepticism has been even sharper for Oracle. Despite beating earnings expectations, the company missed on revenue and failed to provide enough clarity on how it plans to finance its aggressive AI-driven infrastructure expansion. The stock is now down more than 40% from its September record, as investors grow wary of the heavy debt load required to keep pace in the AI arms race.

CoreWeave offers another cautionary tale. The data-center operator, which rents out AI-focused cloud infrastructure, fell 9% on Friday and has lost more than half its value since peaking in June, reflecting concerns about capital intensity and long-term returns.

Taken together, the moves point to a market that is no longer willing to reward AI exposure at any price. The theme that has dominated equities and corporate strategy is still intact, but investors are becoming more selective, focusing less on headline growth and more on margins, cash flow, and balance-sheet risk.

However, some analysts believe the selloff may prove temporary for Broadcom if demand continues to surge as forecast.

President Trump on Thursday signed an executive order that seeks to sharply curtail the power of U.S. states to regulate artificial intelligence, marking one of the most aggressive federal interventions yet in the rapidly expanding AI sector.

The order authorizes the U.S. attorney general to challenge and potentially overturn state laws deemed inconsistent with what the administration calls “the United States’ global A.I. dominance,” placing dozens of existing safety, consumer protection, and transparency measures in legal jeopardy.

Under the order, states that refuse to roll back targeted AI laws could also face financial pressure. Trump directed federal agencies to withhold funds tied to broadband expansion and other infrastructure programs from states that maintain regulations viewed as obstructive. The threat adds a fiscal lever to what is already shaping up as a major constitutional confrontation between federal authority and state police powers.

Trump framed the move as a necessary step to eliminate what he described as a confusing and burdensome regulatory landscape. Speaking in the Oval Office alongside senior officials, including David Sacks, the administration’s AI and crypto czar, Trump argued that innovation could not thrive under a fragmented system of state rules.

“It’s got to be one source,” he said. “You can’t go to 50 different sources.”

He also tied the order directly to geopolitical competition, repeatedly citing the need for the United States to stay ahead of China in artificial intelligence.

The executive action reflects Trump’s broader realignment toward Silicon Valley and the AI industry. Over the past year, his administration has issued multiple orders designed to ease regulatory scrutiny, expand private-sector access to federal data, streamline permitting for data centers and power infrastructure, and loosen restrictions on exporting advanced AI chips. Trump has also publicly praised leading technology executives and elevated Sacks — a venture capitalist with deep ties to the tech sector — into a central policy role with significant influence over AI governance.

However, the order has already sparked widespread bipartisan resistance, with legal experts warning that it may exceed the president’s constitutional authority. States and consumer advocacy groups are expected to challenge the measure in court, arguing that only Congress can preempt state laws on this scale.

Several legal scholars have noted that while federal agencies can set standards in specific domains, a blanket attempt to invalidate state statutes through executive action is likely to face serious judicial scrutiny.

Even some voices aligned with Trump’s ideological camp expressed concern. Wes Hodges, acting director of the Center for Technology and the Human Person at the Heritage Foundation, said that if the administration succeeds in undermining state rules, it has a responsibility to replace them with a robust national framework.

“Doing so before establishing commensurate national protections is a carve-out for Big Tech,” Hodges said, underscoring fears that the order prioritizes speed and scale over public safeguards.

The stakes are high because generative AI systems have moved rapidly from experimental tools to mass-market products. Technologies capable of generating realistic text, voices, images, and video are now embedded across finance, education, healthcare, marketing, and social media. At the same time, documented harms have multiplied, including deepfake political content, financial scams, data misuse, and cases in which chatbots have provided harmful advice to minors.

In the absence of comprehensive federal legislation, states have stepped in aggressively. According to the National Conference of State Legislatures, all 50 states and U.S. territories introduced AI-related bills this year, and 38 states enacted roughly 100 new laws. These measures vary widely but generally aim to impose transparency requirements, restrict certain uses of AI, and hold companies accountable for foreseeable harms.

California adopted one of the most consequential laws, requiring developers of the largest AI models — including OpenAI’s ChatGPT and Google’s Gemini — to conduct safety testing and disclose the results. South Dakota moved to curb election-related manipulation by banning AI-generated deepfake videos in political ads within months of an election. Utah, Illinois, and Nevada passed laws governing AI chatbots used in mental health contexts, mandating user disclosures and limiting how sensitive data can be collected and used.

Child safety has emerged as a particularly active area of state regulation. Several states have passed laws aimed at protecting minors from AI-powered chatbots and algorithm-driven platforms, especially where AI tools simulate emotional support or companionship. Trump’s executive order states that it will not pre-empt child-safety laws, but it does not define how that exemption will be applied, leaving advocates concerned that protections could still be weakened through litigation or narrow interpretations.

“Blocking state laws regulating A.I. is an unacceptable nightmare for parents and anyone who cares about protecting children online,” said Sarah Gardner, chief executive of Heat Initiative, a child-safety advocacy group.

She warned that states have become the primary line of defense as federal action has lagged.

The AI industry, for its part, has mounted an intense lobbying campaign against state-level regulation. Companies argue that complying with dozens of different legal regimes raises costs, slows product development, and discourages startups. Earlier this year, lawmakers attempted to include a ten-year moratorium on state AI laws in a major domestic policy bill, but the proposal was abandoned after strong bipartisan opposition. Venture capitalist Marc Andreessen captured industry sentiment in a social media post last month, calling the state-by-state approach “a startup killer.”

Trump’s order effectively revives that fight through executive authority, raising the prospect of prolonged legal battles that could inject uncertainty into the AI market. While the administration argues that centralization will accelerate innovation and strengthen U.S. competitiveness, critics counter that the absence of binding federal standards leaves consumers, workers, and children exposed at a time when AI systems are becoming more powerful and less transparent.

Beyond domestic policy, the order also signals how Trump views AI as a strategic asset. By tying deregulation to competition with China, the administration is framing AI governance not as a consumer-protection issue but as a national power contest. That framing may resonate with parts of Congress, but it also heightens tensions with states that see immediate local risks from unchecked AI deployment.

As the order moves toward inevitable court challenges, the outcome could reshape the balance of power over technology regulation in the United States. If Trump prevails, states may find their ability to respond quickly to emerging AI harms sharply reduced. If the courts strike the order down, pressure will mount on Congress to finally craft a national AI framework that balances innovation with enforceable safeguards.

Either way, the executive order marks a turning point. It crystallizes a growing divide between federal ambitions to dominate AI globally and state-level efforts to manage its risks locally.