Artificial Intelligencer-The Intel deal that united Trump and Sanders

By Krystal Hu

Aug 27 (Reuters) - (Artificial Intelligencer is published every Wednesday. Think your friend or colleague should know about us? Forward this newsletter to them. They can also subscribe here.)

Two big questions are hanging over the much-anticipated earnings report from Nvidia NVDA.O, the world’s most valuable company: How healthy is the trillion‑dollar AI trade—and where does the world’s most consequential tech relationship between the U.S. and China stand?

First, we'll see if last week's tech stock jitters were a real warning or just noise. As the supplier of the industry’s core hardware, Nvidia is the bellwether.

Then there’s geopolitics. The earnings call is expected to offer the first real read on how an unprecedented revenue‑sharing arrangement with the U.S. government could shape guidance.

When CEO Jensen Huang struck a deal with President Donald Trump to keep selling certain chips to China in exchange for paying Washington 15% of those sales, investors cheered and Nvidia ramped production.

Beijing—despite voracious demand for Nvidia chips—then urged local firms to curb purchases on security grounds. The earnings should hint at how that China slice—and that 15% skim—are impacting its financial outlook.

Nvidia isn’t the only chipmaker in the hot seat. In this week’s issue, we dig into the case for and against partial U.S. government ownership of Intel INTC.O. And with “The Wizard of Oz” opening at the Las Vegas Sphere, we also look at the latest AI capabilities that have Hollywood buzzing. Scroll on.

Email me here or follow me on LinkedIn to share any feedback, and what you want to read about next in AI. 

OUR LATEST REPORTING IN TECH & AI

Exclusive-TikTok owner ByteDance sets valuation at over $330 billion as revenue surpasses Meta

Meta to launch California super PAC backing pro-AI candidates

‘It saved my life.’ The people turning to AI for therapy

OpenAI, Altman sued over ChatGPT's role in California teen's suicide  

Investors worry Trump's Intel deal kicks off era of US industrial policy

Chinese AI startup DeepSeek releases upgraded model with domestic chip support

Meta signs over $10 billion cloud deal with Google, source says

THE DEAL THAT UNITED TRUMP AND SANDERS

When President Trump and Senator Bernie Sanders end up on the same side, you know we’re in new territory.

Both backed Washington’s decision to swap CHIPS Act support for a 9.9% equity stake in Intel. Trump viewed it as a show of industrial muscle, and Sanders as a rare case of taxpayer money buying a real ownership stake.

That convergence sets the stage for a bigger question: how far should the government reach into a “free” market to secure its chip supply chain?

Intel’s own filing nods to the risks: a federal shareholder can complicate overseas sales, trigger foreign subsidy rules, and dilute existing investors. Then, layer on the politics. Trump publicly pressured the CEO before praising him, and the Treasury signaled there’s no plan to own winners like Nvidia while hinting other sectors could be candidates. You can see why markets are uneasy.

The bull case is simple. Chip manufacturing is strategic infrastructure, especially amid the U.S.–China rivalry. Intel’s foundry reboot is capital‑intensive, late and fragile. A government backstop screams “too important to fail,” lowers capital costs and could de‑risk multibillion‑dollar construction in Arizona. Advocates argue that a shareholder‑government can better align national security with supply chains. If Washington wants to secure leading‑edge chip production at home, it helps to have skin in the game.

The bear case is about execution and precedent. Even without a board seat, a 9.9% owner with regulatory power influences decisions on M&A, capital returns and where Intel can sell. That’s a headache when 76% of revenue is international and roughly 29% comes from China. Foreign governments could retaliate or slap new rules on a “state‑backed” Intel.

Investors also worry that government money can't fix fundamental business problems. The company's comeback hinges on its next-generation chipmaking technologies. As 18A, Intel’s node around the 1.8 nm class, still has uncertain manufacturing yields, and its 14A next-generation node lacks anchor customers to commit big volumes, equity from Washington doesn’t fix the technical gap or the sales pipeline.

That could push the government further—owning a stake may not be enough. Washington might need to rally other private companies to become customers and craft incentives for them to use Intel’s chips to help it survive.

Sound familiar? It echoes parts of China’s industrial playbook. Beijing’s 14th Five‑Year Plan makes tech self‑reliance a centerpiece, with semiconductors tied directly to economic and national security. Recent guidance urging domestic tech giants to curb purchases of Nvidia’s top GPUs while promoting local chips (including from Huawei) fits that push.

China’s system can marshal resources, known for concentrating power to get big things done, giving it an edge in scaling certain innovations. But heavy‑handed intervention can also backfire, from misallocated capital to talent flight.

The U.S. now has to strike a delicate balance: support a strategic tech industry without tipping into the kind of state capitalism that markets—and allies—will quickly price in.

CHART OF THE WEEK

In the first rush after ChatGPT, consultants made bank. Executives were desperate for a roadmap, so they bought workshops, audits, and pilot projects to see if AI could do anything useful. That flurry gave the likes of Accenture and Cognizant a lift, and a revenue boom—the “teach me what this is” phase paid well because nobody wanted to miss the moment.

Then, the mood shifted. Once companies locked in the obvious use cases, the money moved to the companies that own the models, the chips and the cloud infrastructure. That’s why the blue lines of the big tech—Amazon, Alphabet, Microsoft and Meta—climb over the past year while the red lines slip. Integrators are left pitching custom projects as AI quietly automates their once-billable work.

AI JARGON YOU NEED TO KNOW

By Kenrick Cai, Technology Correspondent

Some AI heavy hitters are betting that the next frontier is three-dimensional. World models aim to generate 3D simulations of physical space. These models face a new set of technical hurdles beyond the usual text-or-image generation challenges, like enabling real-time interaction with the simulated world. And even if a world is generated correctly, an AI's limited memory makes it difficult to ensure the environment remains physically consistent as it changes over time.

Google’s DeepMind AI division announced the latest version of its world model Genie 3 earlier this month, although it is not publicly available. OpenAI has described its generative video model Sora as a “step towards developing models that can interact with the physical world.” Fei-Fei Li, a pioneering researcher known in the field as the “godmother of AI,” last year raised $230 million for a new startup, World Labs, that is pursuing this research frontier. The technology is too nascent to have real-world impact so far, but successful advances could power everything from generating fantastical video games to providing virtual training grounds for robots.