Lessons from Anousheh Ansari, XPRIZE-winning founders and investor Nancy Pfund on building AI that delivers real-world impact—and lasting success.

Global investment in artificial intelligence is no longer just accelerating—it is fundamentally re-baselining the global economy. According to the International Data Corporation (IDC), annual worldwide spending on AI is projected to reach $632 billion by 2028 , more than doubling in just four years. Meanwhile, PwC estimates AI could contribute more than $15 trillion to the global economy by 2030. Yet for all that capital, a fundamental question remains: how much of it is being used to solve real human problems?

Artificial intelligence is no longer a future bet. It is rapidly becoming the foundation of how industries operate, economies grow and societies function. From climate systems to education and healthcare, AI is reshaping how problems are understood and solved at scale.

For founders, enterprise leaders, investors and educators responsible for preparing the next generation of builders, there is no turning back. The technology has arrived. The opportunity now is to ensure it is used to solve the problems that matter most.

At XPRIZE, where global competitions tackle challenges from climate change to equitable access to education, a consistent pattern emerges: the most successful teams do not begin with technology. They begin with a problem.

To explore this mindset in depth, I spoke with Anousheh Ansari, CEO of the XPRIZE Foundation, about how the world’s most effective innovators approach AI, impact and scale. Watch the full conversation here .

“We look for individuals who are passionate about solving problems… creating a better future for humanity,” Ansari says. “The use of AI is just the means to get to that goal.”

That distinction is more than philosophical. It determines whether an AI solution remains a prototype—or becomes something that works, scales and lasts.

Don’t Start With AI. Start With A Real Human Problem

For all the capital flowing into AI, one mistake still defines too many products: founders start with the technology before fully understanding the problem.

“The end goal is not just profit, it’s impact,” Ansari explains.

Builders who start with human need ask:

  • Who is struggling?
  • What constraint matters most?
  • What outcome improves real lives?

Builders who start with technology ask:

“Start with a real human bottleneck, not with technology.” — Shantanu Agarwal

Shantanu Agarwal, founder and CEO of Mati Carbon and leader of the XPRIZE Carbon Removal-winning team, puts it directly:

His company focused on smallholder farmers facing degraded soil and climate stress. As Agarwal explains, climate solutions scale much faster when they solve an immediate human problem—not just a future atmospheric one. Carbon removal scaled because it solved an urgent need first.

Sooinn Lee, co-founder of Enuma and leader of an award-winning education platform, reached the same conclusion in a very different context:

“AI is more than just connecting a user to a server—you will face real-world hurdles like poor infrastructure and low literacy.”

Her team designed for children in low-resource environments who had never encountered digital tools, ensuring the solution could function autonomously and adapt to entirely new learning contexts.

The best AI companies are built around human urgency—not technical novelty.

Why Most AI Companies Struggle To Scale

Many AI companies fail not because the technology doesn’t work—but because it solves the wrong problem.

Builders focus on what models can do, rather than where real human bottlenecks exist.

The result: impressive systems that fail to deliver real-world impact.

Build For Reality—If It Doesn’t Work In The Field, It Doesn’t Work

Choosing the right problem is only the first step. The next failure point is building solutions that do not survive the real world.

“You may not understand the problem and the setting where the problem manifests well enough… If you’re sitting in your labs… chances are it won’t work.”

In AI, capability does not equal usability.

Thomas Walla, biodiversity scientist and leader of an XPRIZE-winning rainforest monitoring team, saw this firsthand in the Amazon, where environmental conditions forced constant adaptation.

“We had to move forward only with systems we knew would work,” he said.

In practice, those constraints were severe. Heat and humidity destroyed equipment, drones failed and communication systems broke down. Teams were forced to strip away anything that wasn’t field-ready and operate only with systems that could survive real-world conditions.

Similarly, Lee’s platform had to function without reliable infrastructure or prior exposure to digital tools.

  • Design for constraints, not ideal conditions
  • Test where failure is most likely
  • Prioritize usability over capability

Performance in the lab is optional. Performance in the real world is everything.

Scale Matters—Building Scalable AI Solutions For Real-World Impact

Even when a solution works, scaling it remains the real challenge.

This is where many companies fail—not because the technology breaks, but because the business does.

Nancy Pfund, founder and managing partner of DBL Partners and a pioneer in impact investing in AI, explains:

“Mission might help a company get to a first check, but business results determine the rest.”

For AI builders, purpose and profitability must align.

“Founders that change entire industries are clear-eyed from the start. To scale impact, profit and purpose must come together.”

This reality is reinforced by operators scaling solutions globally. Kristal Kaye, CFO and interim CEO of CarbonCure Technologies, an XPRIZE-winning company deploying carbon utilization solutions worldwide, puts it simply:

“Sustainability solutions scale fastest when the financials work… that economic benefit is not a nice-to-have, it’s a must-have.”

Her team’s approach highlights a critical lesson for founders: scaling is not about deploying everywhere at once, but about aligning customer value, economics and adoption in real-world markets.

The real challenge is execution.

“The day-to-day work is where real impact is built, brick by brick,” Pfund explains. “Progress is nonlinear.”

  • Strong ideas are not enough
  • Execution determines impact
  • Resilience determines success

“The AI founders I find most compelling… are building companies that advance progress while meaningfully mitigating factors like societal pushback, environmental stress and cybersecurity.”

The Hardest Problem—Aligning Technology With Humanity

If building human-centered AI that works requires solving real problems, designing for reality and scaling effectively, the next challenge is ensuring these systems align with human values.

AI is not just shaping industries. It is shaping behavior, culture and perception at scale—often faster than institutions can respond.

For founders, educators, investors and policymakers, this creates a new layer of responsibility. It is no longer enough to build systems that perform. They must also be designed with an understanding of how they influence people and society over time.

Across the leaders highlighted here, a consistent pattern emerges:

  • Start with real human needs
  • Build for real-world conditions
  • Scale with discipline and purpose

Increasingly, there is a fourth dimension:

  • Build with awareness of long-term societal impact.

The future of AI will not be defined by what the technology can do.

It will be defined by how deliberately it is applied—and whether it expands opportunity or reinforces existing challenges.

  • Start with a real human problem
  • Build for real-world conditions
  • Test where failure is most likely
  • Align profit with purpose
  • Design for long-term human impact

The question is no longer whether AI will transform the world.

The real question for builders is whether AI is being directed toward the problems that matter most.