Cerebras CEO pushes back on AI doomsday fears

# Cerebras CEO Pushes Back Against AI Doomsday Warnings, Urges Focus on Benefits

Cerebras Systems CEO Andrew Feldman has challenged increasingly dire warnings about artificial intelligence posing existential threats to humanity, arguing that concerns about the technology’s potential dangers should not overshadow its opportunities to improve healthcare and accelerate scientific research.

Speaking at TechCrunch Disrupt 2026, Feldman criticized the increasingly negative narrative surrounding AI and suggested that industry leaders should communicate its risks alongside the potential benefits. According to TechCrunch, his remarks addressed concerns ranging from AI systems escaping containment to predictions that advanced AI could threaten humanity before the end of the decade.

Feldman argued that discussions about AI should give greater attention to the advances it could enable, particularly in medicine. He pointed to the possibility that AI could help make cancer significantly more treatable, saying the public should hear more about such potential breakthroughs rather than focus predominantly on worst-case scenarios.

Cerebras develops specialized chips designed for AI computing, placing the company within an industry whose growth is closely tied to the continued adoption of advanced AI systems. TechCrunch reported that Cerebras customers include OpenAI, Block, Figma, pharmaceutical company GSK and Lovable.

The company's commercial interests provide important context for Feldman’s position, as greater AI adoption can create demand for the computing infrastructure and hardware that Cerebras supplies. Nevertheless, he acknowledged that the technology’s capabilities also create serious challenges, especially when powerful systems are used by malicious actors.

“I don’t think there are powerful technologies that don’t come with the challenge of ‘what happens when powerful technologies are in the hands of bad people?’ I think you can’t make one and not have the other,” Feldman said during his interview at the event, according to TechCrunch.

He maintained that acknowledging the risks should not prevent the industry from developing AI, but should instead encourage responsible innovation and appropriate oversight.

“I think what we have to say is that this technology is powerful,” Feldman said. “We have to be thoughtful about it. We have to regulate it reasonably.”

His comments come amid a wider debate over how governments, technology companies and researchers should manage increasingly capable AI systems. While some industry figures have warned of potentially catastrophic consequences, others have emphasized the importance of balancing safety measures with continued research and development.

## Data Center Expansion Creates Local Challenges

Feldman also discussed the public opposition facing AI infrastructure projects, linking some of the resistance to the way technology companies have handled the expansion of data centers.

As AI models become more sophisticated and widely used, demand for computing capacity has increased, encouraging companies to build new facilities and expand existing ones. These projects can raise concerns among nearby residents about electricity consumption, infrastructure demands and the effects of construction on local communities.

Feldman acknowledged that technology companies had not always done enough to engage residents affected by data center development, particularly in the United States.

“As an industry, we weren’t super good for the past few years at working with local communities, particularly in the U.S.,” he said, according to TechCrunch.

He suggested that companies could learn from established energy industries, including oil and gas, which have experience working with communities around major infrastructure projects. His comments underscored the importance of building local relationships as AI companies seek approval and support for facilities that require substantial resources.

Feldman also contrasted resistance to some U.S. data center projects with efforts in countries including Brazil, Canada, Finland, Iceland, Norway, Mexico and Sweden to attract computing infrastructure investment. His remarks highlighted how public engagement and local conditions can influence where companies choose to establish facilities.

The debate over data centers extends beyond the technology sector, touching on electricity supply, infrastructure planning and the distribution of the economic benefits and costs associated with AI expansion. As investment continues, companies face pressure to demonstrate how their projects will affect the communities hosting them.

## Efficiency a Shared Responsibility for Hardware and AI Developers

Feldman also addressed whether improvements in AI efficiency could reduce demand for specialized computing hardware, including chips developed by Cerebras and competitors such as Nvidia.

He argued that hardware manufacturers and AI developers should work together to improve performance and efficiency rather than treat progress in one area as a threat to the other. More efficient chips and better-designed models, he suggested, can complement each other in delivering improved results.

“We can do our job as hardware vendors to make our compute more efficient, and then the model makers can do improved work in making their models more efficient to achieve the same goals,” he said, according to TechCrunch.

“All of this has to come together to make meaningful strides forward,” he added.

The comments reflect a broader challenge facing the AI industry: delivering increasingly capable systems while managing computing costs, energy requirements and safety concerns. Hardware companies are working to improve the efficiency of AI processing, while developers are seeking ways to achieve better results with fewer computational resources.

Feldman’s position emphasizes a future in which technological progress, reasonable regulation and greater attention to community concerns advance together. Although disagreements remain over the scale of AI’s potential risks, his remarks argue that the public discussion should also consider its possible contributions to medicine, scientific discovery and other fields.

The debate is likely to continue as AI capabilities expand and investment in the infrastructure supporting them grows. For Feldman, the challenge is not simply to recognize the technology’s power, but to ensure that its development is accompanied by safeguards, efficient computing and more constructive engagement with the public.