Building AI Responsibly: Schwab Network Features Column CEO Megan Villanueva

Column CEO Megan Villanueva joined Schwab Network to discuss NVIDIA’s $12.9 billion acquisition of Hugging Face — and why the next phase of AI growth needs a much bigger conversation about efficiency, security and the physical resources behind it.

NVIDIA’s acquisition of Hugging Face brings together the dominant provider of AI computing hardware with one of the world’s largest platforms for open AI models, datasets and developer tools. NVIDIA CEO Jensen Huang has framed the deal as an opportunity for Hugging Face to “serve the global AI community at unprecedented scale,” while making AI more open, capable and accessible.

On Schwab Network’s Trading 360, hosted by Marley Kayden, Column CEO Megan Villanueva joined Ivan Feinseth of Tigress Financial Partners to discuss what that next phase of growth could mean — and what the AI industry may be overlooking as it races to scale.

Scale is only half of the conversation

There is plenty to be excited about in the move toward open models. Hugging Face gives developers and organizations more flexibility to find and customize models for specific use cases rather than relying solely on the largest general-purpose systems. Huang himself has argued that open models allow organizations to “match the right model to the right job.”

That could be an important step toward more efficient AI.

But as Villanueva emphasized on Schwab Network, the industry still isn't talking nearly enough about efficiency.

AI leaders talk constantly about expanding access, increasing capability and scaling infrastructure. The harder question is how we achieve the same — or better — outcomes while requiring less compute, less energy and fewer physical resources.

AI does not scale in the abstract. More compute means more data-center infrastructure, greater electricity demand, cooling requirements and, ultimately, greater demands on the communities supplying the land, water, power and infrastructure behind AI growth.

That makes the local process around data-center development increasingly important, too. As new facilities move through zoning, permitting and public-hearing processes, residents need clear information about what is being proposed and meaningful opportunities to weigh in.

We recently explored how public notice and local newspapers help communities stay informed about data-center development and participate before major decisions are made.

Villanueva raises a central question: If AI is going to scale at an unprecedented rate, how do we make sure that doesn’t create an unprecedented burden on the communities supporting that growth?

Openness and efficiency have to grow together

NVIDIA’s investment in open models is encouraging. Open models can allow organizations to choose smaller, more specialized systems that are suited to the task at hand rather than using the largest model available for everything.

But greater accessibility can also mean dramatically greater usage.

If AI adoption grows exponentially without corresponding improvements in efficiency, making AI more accessible could still translate into greater overall demand for compute, electricity and data-center infrastructure.

Openness matters. Accessibility matters. But efficiency needs to be part of the conversation too, particularly when the industry is making acquisitions and investments at this scale.

The next phase of AI shouldn’t just be about how much more we can build. It should be about how intelligently, and responsibly, we can build it.

Scale also raises new security questions

Efficiency isn’t the only challenge growing alongside AI capabilities.

In July, Hugging Face disclosed a significant security incident involving an autonomous AI agent system that accessed portions of its production infrastructure. The company described it as different from previous attacks precisely because AI powered the intrusion end-to-end — while AI tools were also critical to detecting and understanding it.

That incident underscores another version of the same principle: “move fast and scale” cannot be the only measure of progress. As AI systems become more capable, widely available and autonomous, the governance, testing, security and human judgment surrounding them have to advance too.

The NVIDIA–Hugging Face deal is about much more than one acquisition. NVIDIA is expanding beyond its dominant position in AI hardware and deeper into the models, tools and platforms developers use to build AI. Hugging Face, meanwhile, has become a central piece of the open-model ecosystem, with more than 18 million developers and more than 200,000 companies using its platform.

As that ecosystem grows, we shouldn’t just be asking how big can AI become? but also how do we make that growth more efficient, secure and sustainable for the communities making it possible?