Artificial intelligence (AI) has never been more popular, or unpopular, depending on who you ask.

Nvidia, with its $5.2 trillion market cap, is the company that has undoubtedly benefited most from the AI boom, so it’s wholly unsurprising that the GPU giant is doing its best to try and win over the industry’s naysayers.

In the US, the Trump administration is largely supportive of AI (or super intelligence – as the President recently dubbed it) and opposes heavy-handed government regulation, instead favoring a self-regulatory approach which it claims breeds innovation. By comparison, the European Union has a long history of passing legislation with the hopes of keeping the technology sector in check.

In Europe, AI safety is a primary issue of concern, particularly in the face of recent reports about AI agents from OpenAI and Anthropic going rogue. But, as is also the case across the pond, while governments are largely supportive of continued data center build-out to support sovereign AI efforts, local communities are fiercely opposed, with concerns around energy and water resources often cited as the primary issue.

So how does a company whose profit margins are tied to the ever-increasing data center construction try to make friends and influence people in a region where its corporate agenda is increasingly at odds with both government and society?

At GTC Berlin last week, that’s exactly what Nvidia tried to do … to varying degrees of success.

When it comes to AI, Nvidia CEO Jensen Huang told reporters there are three main objections at present: Energy, safety, and capital. Therefore, making a number of sustainability- and safety-focused announcements at GTC Berlin certainly seemed like smart positioning from Nvidia, starting with claims related to water and energy savings that it said its DSX platform can make.

Described by Nvidia as “responsible by design,” the chip giant (which now prefers to be known as a full-stack platform provider) said its DSX Platform will help make facilities more water- and power-efficient, allowing customers to build AI data centers that minimize the amount of water consumed by cooling technologies by up to 100 percent, whilst also utilizing up to 30 percent less power for the same output.

While helping data center operators save 54 million liters (14.27 million gallons) of water a day sounds like an impressive feat, it should be noted that figure relates to a 2GW data center deployment, something which doesn’t yet exist, despite many gigawatt-scale campuses having been announced around the world. Nvidia has not provided stats for how much water the platform can save for data centers operating at mere-megawatt scale.

The company also unveiled a reference architecture at the conference based on its OpenShell software. The full-stack offering enforces policy-based security, network, and privacy guardrails for AI agents at the system layer, and now covers Vera Rubin infrastructure for running enterprise workloads, Vera CPUs to operate agent harnesses, and BlueField DPUs for running security software frameworks. It also encompasses Edge-based systems like DGX Spark and Station for agents operating across retail, telecom, and robotics.

OpenShell covers models from AI labs and builders including Anthropic, Cohere, and OpenAI, although the latter was notably absent from both Nvidia’s Open Agent Safety group and the Open Secure AI Alliance.

Nvidia might have spent two days saying all the right things to a European market that has just lived through a summer of record-breaking heatwaves and has long-sought to regulate Big Tech to prevent public harms; however, when Huang was subject to a slightly combative 90 minutes of questioning by a room full of European journalists, that public messaging felt a little hollow.

When asked about the power bottleneck Europe is currently facing and whether Nvidia is looking to play a bigger role in solving the energy problem, while Huang did acknowledge the apparent benefits of its DSX platform, he seemed more keen to make sure the room understood that pulling back on data center build-out was definitely not the solution.

“If anybody believes in a sustainable future, the answer is use more energy,” the CEO said. “If there’s no demand for energy, why would anyone invest in the energy industry?”

Apparently, if the world decides to stop deploying GPUs at scale, then everyone who works in Green Tech would presumably give up and go home – although perhaps not, because “AI is creating jobs in the energy sector … give me an example of an energy company today who doesn’t think ‘AI is fantastic!’”

Given the OpenShell announcements, AI safety and regulation were another topic on which, understandably, Huang faced scrutiny.

As the industry continues to drink the agentic AI Kool-Aid, OpenShell is a tool that feels increasingly necessary, but also one that certain segments of the industry clearly hope governments (particularly in Europe) will defer to instead of pursuing heavy-handed regulation.

Huang made his opposition to hypothetical legally enforceable AI regulatory frameworks very clear, scoffing at the mere concept and labeling such attempts “ridiculous,” arguing that companies should be able to make their own judgments and act responsibly when it comes to launching products that could have the potential to cause harm.

“If you know that your product is delivering harm to society, stop right away,” he sniped. “And if you believe that you need someone to force you to do it, then you can go and advocate for someone to force you to stop, but you don’t have to advocate for someone to force everybody to stop.

“Ask every single car company in Europe, if your car is not ready to be released, do you need society to tell you not to release it?” Huang added, failing to note that the European auto industry is one of the most heavily regulated in the world – just ask Elon Musk why you never see Tesla Cybertrucks on European highways.

Regarding the final AI objection, capital, several reporters attempted to dig deeper into some of Nvidia’s recent funding announcements. While accusations of “circular financing” weren’t directly thrown at Huang, he was asked how far Nvidia might be prepared to go to back neoclouds such as CoreWeave and Nebius, to which Nvidia has provided equity investments, loans, credit support, and backstops, enabling them to purchase more GPUs.

“We invested in them, that’s different from backing them,” Huang pushed back. “We invested in them because I think it’s a good investment and I can prove it … their compute demand is incredible … and completely sold out.”

And what does sold-out compute demand mean? That these companies need to raise more capital to buy additional compute infrastructure, ad nauseam (and hopefully, for Huang, from Nvidia).

At this point, it would be remiss not to mention that Musk took to X this week, claiming that “as of the past few weeks, debt markets have almost entirely shut off access to money for any company that doesn’t have substantial revenue independent of AI data centers.

“That means the 200+ neoclouds are going to see guppy-level death and consolidation. Some will make it through, but not many.”

At least Huang had some words of wisdom for neocloud Nscale – which hasn’t been handed a death sentence but, this month alone, has seen Microsoft withdraw from the Essex data center the cloud company has been developing, and six-month delays reported at its Glomfjord, Norway, site.

When asked if he had any advice for the AI cloud firm, he suggested that there’s no replacement for working harder, and that Nscale should follow his lead and sleep less. The sincere answer the room was waiting for never came.

Huang ended the Q&A by reassuring the gathered journalists that questions around energy, AI safety, and capital were “not unique to Europe” and that these concerns were important to every country. While that might be the case, you couldn’t help but feel that the Nvidia boss is rather hoping that the world stops asking about them.