On Tuesday, President Donald Trump brought many of the most powerful people in AI to the White Home, together with Nvidia CEO Jensen Huang, Meta CEO Mark Zuckerberg, Google CEO Sundar Pichai, and Anthropic CEO Dario Amodei.
At a excessive stage, most need some model of the identical factor: extra information middle capability, sooner allowing, extra federal spending on AI, and a light-weight (or nonexistent) regulatory contact. Their priorities overlap, generally significantly, however the firms’ totally different companies give them totally different stakes within the particulars. Some, for instance, are pushing the federal authorities to impose new guidelines on essentially the most highly effective AI methods, whereas others are preventing restrictions on chip exports.
We don’t know the whole lot the executives requested Trump for in non-public, however their firms have typically been fairly specific about their priorities in Washington, D.C., and this was an opportunity to attempt to affect the president to their particular positions. Right here’s what every had at stake.
Anthropic needs the federal government to control frontier AI
For a lot of the previous 12 months, Anthropic brass has argued that essentially the most superior AI makers want extra authorities oversight—which might possible imply imposing new necessities on Anthropic itself.
The corporate’s proposed “regulatory ladder” would make the principles more durable as AI methods change into extra succesful, ultimately bringing in exterior testing and incident reporting. Anthropic says its Superior AI Framework can perform as a road map for policymakers, together with giving governments authority to dam or stymie high-risk deployments.
Dario Amodei additionally got here into Tuesday’s assembly after months of preventing with the Trump administration over the Pentagon’s use of Claude. In March, after Anthropic refused to drop restrictions on mass home surveillance and totally autonomous weapons, the Pentagon designated the company a supply-chain risk. Anthropic challenged the transfer, however a federal appeals court docket upheld the designation in September. Nonetheless, relations look like thawing, as Trump hosted Amodei for a private dinner a couple of days earlier than the bigger White Home gathering.
Nvidia needs Washington to not do something that stops AI’s demand for extra chips
As a number one chipmaker, Nvidia advantages because the AI business builds extra fashions and information facilities. It’s no shock, then, that its CEO says governments ought to regulate actual harms as an alternative of attempting to foretell each attainable threat.
At a G20 event in September, Jensen Huang mentioned policymakers ought to concentrate on “sensible and precise hurt” slightly than “theoretical and hypothetical hurt.” The Wall Road Journal reported that Huang confronted Amodei on the White Home assembly over his public warnings about AI threat.
Nvidia additionally has billions using on U.S. controls on exports of superior AI chips to China. The federal government has repeatedly restricted which Nvidia processors will be bought there with out a license. In 2025, these restrictions left Nvidia with $4.5 billion in charges associated to H20 chips it might not freely promote in China. Newer guidelines have allowed the corporate to renew some gross sales, however Nvidia says it has been in a position to ship solely a fraction of the H200 chips permitted for export. The corporate warns that dropping entry to China provides native opponents extra room to develop.
Meta needs to guard open AI
Meta, a pioneer in open-source AI (by which mannequin weights are launched so builders can run and construct on them), has a powerful stake in whether or not the federal authorities attracts a regulatory distinction between closed frontier fashions and open ones. In spite of everything, restrictions on how superior fashions will be launched or distributed might lower immediately towards the technique Meta has spent years pursuing.
In an August essay, Mark Zuckerberg claimed that restrictions on entry to main open-source fashions might focus AI in fewer arms. And, he argued, the U.S. wants to steer in open-source AI as a result of these fashions are prone to change into extra fashionable globally, and thus signify a chance to increase delicate energy.
The Meta CEO additionally warned towards guidelines that slim what information American builders can use for coaching or prohibit mannequin “distillation,” the apply of coaching one AI system on the outputs of one other. He says these limits might go away American fashions at a drawback to Chinese language opponents.
Google needs one federal rulebook
Within the checklist of recommendations Google printed final 12 months for the Trump administration’s AI Motion Plan, the tech large referred to as for preserving entry to information for mannequin coaching, increasing power provides, streamlining federal AI procurement, and avoiding export controls that shut American firms out of overseas markets. It has additionally pushed the federal authorities to preempt what it calls a “chaotic patchwork” of state guidelines governing frontier AI.
CEO Sundar Pichai has made the same case, arguing on the 2025 AI Motion Summit that governments ought to “deal with dangers, with out stymying innovation,” and keep away from fragmented regulatory regimes.
Google has extra publicity to federal AI coverage than most of its rivals as a result of it sits across nearly every part of the market: It builds frontier fashions, sells cloud computing, operates huge information facilities, sells AI instruments to companies and governments, and runs the world’s dominant search engine.
OpenAI needs federal security guidelines, however aimed on the greatest labs
Like Anthropic, OpenAI needs federal guidelines for firms constructing essentially the most highly effective AI methods. However Anthropic has gone additional in what it needs regulators to have the ability to really do.
Final month, OpenAI really useful Congress enact obligatory, capability-based nationwide AI safety requirements that embody frequent testing requirements, unbiased assessments, and obligatory reporting of AI-related safety breaches. The corporate argues that the strongest necessities ought to apply to the small variety of firms constructing essentially the most succesful methods, slightly than startups and smaller builders. (Anthropic, by comparability, has referred to as for regulators to have the ability to block harmful deployments and levy civil penalties.)
On the similar time, OpenAI wants an enormous buildout of computing and power infrastructure to continue to grow, giving it cause to assist tighter guidelines for frontier fashions whereas pushing for fewer obstacles to constructing the info facilities and energy methods behind them.
SpaceXAI needs Washington to make use of Grok
SpaceXAI needs to change into the go-to AI equipment for the federal authorities. Certainly, Elon Musk’s firm already has a government-wide deal that lets every collaborating federal division, company, or bureau present Grok to its staff essentially free of charge via March 2027, with SpaceXAI engineers available to behave as IT assist. Musk said the objective was to “quickly deploy AI all through the federal government.”
The Pentagon is a part of that effort. In 2025, SpaceXAI acquired one of four AI awards with a $200 million ceiling to develop instruments for national-security missions, and Grok has since been added to the Pentagon’s AI platform for army and civilian staff. All of which is to say: Musk has a transparent stake in turning federal approval of Grok into additional federal adoption.
Microsoft wants the federal government to assist clear up AI’s infrastructure drawback
Microsoft’s pursuits more and more run via the bodily infrastructure required to maintain its AI enterprise rising. The corporate’s “Community-First AI Infrastructure” plan requires working with utilities so as to add electrical energy and grid infrastructure the place its information facilities want it, whereas promising that residential prospects received’t bear the ensuing prices. Microsoft has additionally dedicated to masking wanted infrastructure upgrades and lowering its demand on native water provides.
For CEO Satya Nadella, federal AI coverage subsequently reaches properly past mannequin regulation, significantly as Washington appears for methods to speed up the large energy and infrastructure buildout the business says it wants.
Palantir needs federal companies to make use of extra AI
Palantir has spent years promoting software program to the federal authorities, so its pursuits are unusually simple.
In its response to the White House AI Action Plan, Palantir referred to as for modernizing how federal companies purchase and deploy AI and even really useful that each company full a brand new flagship AI undertaking inside 9 months of the plan’s publication.
For Palantir, Washington shouldn’t be merely writing the principles governing AI. It is without doubt one of the firm’s most essential potential customers of the expertise.
AMD has billions using on Trump’s chip insurance policies
AMD shares a few of Nvidia’s issues about export controls, however its market place makes federal coverage particularly essential. The corporate is already a part of Washington’s effort to construct extra home computing capability. AMD and the Division of Vitality are working collectively on two new methods at Oak Ridge Nationwide Laboratory, together with an AI supercomputer that the corporate final 12 months described as a part of an “open American AI stack.”
U.S. export controls on superior AI chips have additionally hit AMD immediately. The federal government started imposing the current generation of restrictions on gross sales to China in 2022, arguing that the chips can assist superior AI, supercomputing, and army purposes, and has repeatedly tightened them since.
In April 2025, the Trump administration added a license requirement masking AMD’s MI308 AI chip. AMD says that restriction finally led to about $440 million in inventory and related charges in 2025. The corporate has since acquired licenses to promote some further AI chips in China, however says export controls might nonetheless harm its income and competitiveness.
Bonus: Why was Jeff Bezos there?
Amazon, in fact, has lots using on federal AI coverage. The corporate has mentioned the U.S. must expand energy and infrastructure to accommodate information middle development, and has backed extra constant state and federal AI requirements. However Amazon founder Jeff Bezos is a curious presence on the attendee checklist as a result of he hasn’t really run the corporate since 2021.
Why did Bezos get the invite as an alternative of Amazon CEO Andy Jassy or Amazon Net Providers CEO Matt Garman? Amazon could have merely introduced out the most important (that’s, richest) gun it needed to the present. Plus, Bezos stays Amazon’s government chair and a serious shareholder, so he nonetheless has loads of cash using on how the corporate fares.
