Elon Musk, OpenAI, and the global AI power struggle

Artificial intelligence isn’t just about clever chatbots anymore. It’s now at the center of billion-dollar lawsuits, geopolitical chip wars, and a growing public backlash. Recent events in the US and China show that the real fight over AI is about power: who controls the models, who owns the chips, and who benefits from the technology.

Musk vs Altman: when a “non-profit” becomes a trillion-dollar giant

One of the week’s biggest AI stories played out in a courtroom in Oakland, California. Elon Musk accused Sam Altman and OpenAI of betraying the company’s original mission. Musk says he donated nearly $38 million on the understanding that OpenAI would be a non-profit, building AI “for the good of all humanity.” Instead, OpenAI created a commercial arm, partnered deeply with Microsoft, and grew into a company now valued around a trillion dollars.

Musk’s lawsuit tried to unwind that commercial structure, remove Altman from the board, and claim around $150 billion in damages. The jury ruled against him—not because they decided OpenAI had definitely behaved perfectly, but because Musk waited too long to sue. The statute of limitations had passed.

For OpenAI and Microsoft, this was a huge win. It clears a major cloud of uncertainty over their partnership and paves the way for massive data center expansion that could cost hundreds of billions of dollars. For Musk, it’s a public loss, but not a fatal one. He still controls Tesla, SpaceX, and xAI, and remains one of the world’s richest and most influential tech figures.

Are AI giants really “for humanity” – or just for profit?

The case also exposed a deeper truth: AI labs are not charities. However inspiring their mission statements, they are locked in fierce competition for money, market share, and cultural dominance.

Today’s leading labs aren’t just racing to build the most capable model. They’re also fighting to become the default verb in everyday life. Just as “Google it” became shorthand for web search, “ChatGPT it” is becoming shorthand for using generative AI. Whoever wins that mindshare can lock in users, enterprise contracts, and influence over how AI is used.

That competition is visible even at global AI summits. Executives talk about safety and cooperation on stage, but behind the scenes they’re battling over benchmarks, enterprise deals, and developer ecosystems. The Musk–Altman clash is just one public flare-up in a much larger struggle over who steers the most powerful AI systems on Earth.

If you want a deeper look at Musk’s own AI ambitions, including his Grok model and xAI’s internal turmoil, check out our coverage of Grok 4.3 and xAI’s model strategy and why xAI has been bleeding talent.

Anthropic, Mythos, and the strange new politics of AI safety

In US politics, AI has created some surprising alliances and contradictions. Donald Trump has positioned himself as a champion of aggressive commercialization and American tech dominance. But parts of his own MAGA base are now calling for tougher AI safety measures.

One big reason: Anthropic, a leading AI safety–focused company, privately warned that a system called Mythos could be powerful enough to expose critical vulnerabilities across global infrastructure—banking, utilities, weapons systems, and more—before organizations have time to fix them. Anthropic convened a closed-door meeting of major tech firms to discuss the risks.

That move highlighted two ironies. First, Anthropic was effectively doing the job regulators should be doing: coordinating industry around a potentially catastrophic risk. Second, the Trump administration had previously labeled Anthropic a “supply chain risk,” even as the company was trying to improve national and global security.

Critics also note that by inviting only a select group of “top table” companies into that discussion, Anthropic and its peers may be reinforcing a two-tier AI world. The biggest players get early access to vulnerability information and can patch faster, while smaller firms and critical infrastructure providers risk being left behind.

Public sentiment is turning sharply against AI

While tech leaders and politicians argue about safety and strategy, public opinion is moving in a stark direction. A recent Axios poll found that:

• Over 70% of Americans think AI is moving too fast.
• Negative views of AI have doubled in just three years.
• Only 18% of young people say they feel hopeful about AI.

If AI were a political candidate, Axios concluded, it would be “losing in a landslide.”

Part of the problem is messaging. Many AI CEOs publish long essays predicting that AI will wipe out most white-collar jobs within a couple of years—and then seem surprised when people react with fear and anger. In countries like the US, where graduates can leave university with six-figure debt, the idea that AI will erase entry-level jobs feels like an existential threat, not an exciting innovation.

Why students are literally booing AI

That anxiety is now visible in public. At a recent university graduation in Florida, a commencement speaker described AI as the “next industrial revolution.” The crowd booed. She went on to acknowledge their fears: that the future feels pre-written, that machines are coming for their jobs, that they’re inheriting economic, political, and climate crises they didn’t create.

For many graduates, AI isn’t just another tool. It’s a symbol of a system that may not have room for them. Companies are already using AI for research, analysis, and junior-level tasks—the very roles new graduates used to fill. When the cost of hiring inexperienced workers is high, and AI tools are getting cheaper and more capable, employers have a strong incentive to automate those entry points.

Ironically, the same students booing AI are often heavy users of it for homework, projects, and creative work. They know how powerful it is—which is exactly why they’re worried about what it will do to their job prospects.

What does “good AI” actually look like?

So how do we reconcile AI optimism with policies that often prioritize growth over people? One answer is to get specific about where AI is genuinely beneficial and where it’s harmful or destabilizing.

In some sectors, AI is clearly a net positive. Healthcare is a good example: AI systems are helping doctors read scans, predict disease risks, and manage workflows. Because healthcare is highly regulated, AI tools must meet strict safety and accountability standards before they’re deployed. Few patients want those systems rolled back; if anything, they want them improved and expanded.

The harder questions arise when AI is used to cut jobs, supercharge misinformation, or amplify mental health issues and social division. If AI makes it harder for young people to find stable work, buy homes, or plan families, it undermines social trust. If it floods information spaces with convincing fakes, it undermines democracy.

That’s where the “social contract” comes in. If AI-driven productivity gains flow almost entirely to a handful of companies and investors, while everyone else faces more precarity, backlash is inevitable. Policymakers now have to decide whether existing laws are enough to rebalance that equation—or whether new regulation is needed to ensure AI development actually serves the wider public.

Why AI labs need public support to keep scaling

There’s another practical reason AI companies need to win back public trust: compute. Training and running state-of-the-art models requires enormous data centers, which demand land, electricity, water, and grid capacity. Those facilities have to be built in real communities.

As public sentiment sours, local resistance is growing. Planning applications for new data centers are being challenged or blocked. Residents worry about noise, energy use, water consumption, and the broader question: why should their town bear the costs so a distant tech giant can profit?

If that resistance hardens, it could become a real bottleneck on AI progress. Even the most advanced model architecture is useless without enough compute to train and deploy it at scale.

Jensen Huang, AI optimism, and a different message to graduates

Not every AI leader is getting booed off stage. Nvidia CEO Jensen Huang also gave a commencement speech recently, and his reception was much warmer. The difference was tone and framing.

Huang talked about AI as an opportunity and a shift in how work will be organized, rather than as a blunt threat to entry-level jobs. He encouraged students to consider careers like plumbing and electrical work—skilled trades that are hard to automate and will be in high demand as AI-driven infrastructure expands.

That doesn’t erase the disruption AI will cause, but it does acknowledge a key reality: the future of work will be different, not simply nonexistent. Graduates want to hear about concrete paths forward, not just abstract promises of “efficiency” and “innovation.”

The US–China chip war: who controls the compute?

While legal and cultural battles play out in the US, another fight is unfolding thousands of miles away: the struggle over AI chips. Advanced GPUs are the fuel of modern AI, and right now Nvidia dominates that market.

The US government has tried to slow China’s AI progress by restricting access to Nvidia’s most powerful chips. Recently, it approved limited sales of a slightly cut-down model, the H200, to ten major Chinese firms including Alibaba, Tencent, ByteDance, JD.com, and Lenovo.

But there’s a twist: those companies haven’t actually taken delivery. Beijing appears to be saying “no thanks,” and instead is doubling down on building a fully domestic AI stack—chips, software, cloud infrastructure, and supply chains that don’t depend on US technology.

From Washington’s perspective, export controls are meant to preserve a US lead in AI by limiting China’s access to top-tier compute. From Beijing’s perspective, they’re a loud warning: any reliance on foreign tech is a strategic vulnerability that can be weaponized. That lesson was driven home when sanctions hit Huawei, once a global leader in AI infrastructure and smartphone chips.

How China is adapting: less compute, more efficiency

Export controls haven’t stopped China’s AI ambitions; they’ve changed the way Chinese companies innovate. When you can’t rely on the latest Nvidia hardware, you’re forced to get more out of what you have.

That’s where models like DeepSeek come in. DeepSeek has shown that with clever optimization and algorithmic improvements, you can achieve competitive performance using less powerful hardware. Instead of brute-forcing everything with massive GPU clusters, you squeeze more efficiency out of every chip.

China has a long track record of innovating under constraints, and AI is no exception. As domestic chip platforms like Huawei’s Ascend improve, they’re becoming viable alternatives to Nvidia’s CUDA ecosystem. DeepSeek V4, for example, already runs on Ascend hardware rather than Nvidia GPUs—a symbolic and strategic milestone.

For a broader look at how DeepSeek V4 stacks up against US models like GPT-5.5, and what that means for the compute race, see our breakdown of GPT-5.5 vs DeepSeek V4 and the growing compute war.

Nvidia’s dilemma: sell to China or stay the undisputed king?

Nvidia sits at the center of this storm. On one hand, China is a massive potential market. If Nvidia could sell its most advanced chips there without restrictions, it would make enormous profits. On the other hand, US policymakers want to maintain a technological edge by limiting China’s access to top-tier GPUs.

For now, it’s a “win-win” of sorts for Nvidia. If it can’t sell freely into China, it still gets to market its chips as the world’s best, and US and allied customers have little choice but to buy American. The real threat to Nvidia is the moment when Chinese chips reach parity—or surpass them. At that point, China’s domestic ecosystem could break Nvidia’s dominance.

And that dominance is real today: Nvidia’s CUDA software stack underpins roughly 85% of the AI ecosystem. But as more leading models run on non-Nvidia hardware, that grip could loosen. The race is no longer just about who has the single best chip; it’s about who can build the most robust, independent, and scalable AI system end-to-end.

The future of AI: power, trust, and hard choices

Put together, these stories paint a clear picture. AI’s future will be shaped by three intertwined battles:

• Corporate power: Who controls the leading models and platforms, and are they truly aligned with the public interest or just shareholder value?
• Public trust: Will people accept AI in their workplaces, communities, and democracies—or will backlash slow deployment and reshape regulation?
• Geopolitics and chips: Can the US keep its lead in AI hardware and software, or will China’s push for self-reliance create a parallel, competing ecosystem?

AI is not going away. The question now is who gets to steer it, under what rules, and with what safeguards. Whether you’re a developer, policymaker, or just someone trying to navigate your career in an AI-saturated world, those decisions will affect you far more than any single model release.