Mr. Casey participated alongside industry leaders including CEOs from Anthropic, Google / Alphabet, AWS, Arm, Blackstone, Blackrock and senior White House officials including President Donald Trump.
At the high-profile summit, he brought his unique perspective on advancing energy efficiency within data center server architecture to enable expanded AI processing capabilities.
Amber Semiconductor, Inc. (AmberSemi™), a fabless semiconductor company pioneering the next generation of power conversion, control, and protection technologies, announced that CEO Thar Casey attended the Strategy Summit for Energy and Innovation for AI Data Centers, an exclusive, high-profile, invitation only gathering hosted by Senator McCormick of Pennsylvania.
Also Read: CIO Influence Interview with Dipto Chakravarty, Chief Product and Technology Officer at Black Duck
The summit, which took place Tuesday, July 15 at Carnegie Mellon University in Pittsburgh, PA, convened global leaders across AI, energy, infrastructure, investment, and government. Attendees included prominent executive leaders, such as Anthropic CEO Dario Amodei, Blackstone President and CEO Jon Gray, Bechtel Chairman & CEO Brendan Bechtel, Alphabet and Google president and Chief Investment Officer Ruth Porat, Arm CEO Rene Haas, Amazon Web Services CEO Matt Garman, among others.
The event included President Trump and key members of his administration, such as David Sacks the White House’s AI and Crypto Czar, Chris Wright the Secretary of Energy, and other executive branch department leads.
AmberSemi’s participation reflects the company’s emerging leadership in addressing one of the most urgent challenges facing AI infrastructure : efficient, high-density power delivery. Thar Casey brought his unique perspective on how next-generation power architectures can enable scalable AI deployment without requiring massive new energy infrastructure.
“It’s a big honor, as a disruptive young company, to be invited and contribute alongside such an esteemed list of top global company heads in AI, data centers, and energy, as well as high level government officials,” said Thar Casey AmberSemi CEO. “It was a great event and kudos to Senator McCormick and the President for making it possible. The discussions were very high energy and outlined clear strategies for the success of AI, Data Centers and energy leadership – not just for Pennsylvania, but for the country as a whole.”
He continued, “AmberSemi represented a unique perspective, one that addresses the growing energy demands of AI not solely by increasing energy supply but by tackling inefficiencies from within the data center itself. From the server rack to the motherboard and down to the AI chip, we focus on the critical areas where energy is most often lost, and power delivery remains inefficient.”
Mr. Casey continued, “Power is the key constraint to scaling AI processing, and our solution directly addresses that challenge. We deliver significant efficiency improvements at the data center server level, enabling higher power density and more efficient delivery of power directly to the AI chip. This isn’t just about energy savings—it’s about unlocking the ability to expand AI capabilities within our current infrastructure, without being constrained by the requirement for new power plants. Our technology makes it possible to do more with the existing infrastructure,” he concluded.
AmberSemi is fundamentally rearchitecting how power is delivered to AI chips, addressing one of the most pressing challenges in data center scalability. By collapsing the traditional power delivery architecture, Amber relocates the power supply from the server rack directly to the backside of the motherboard—positioned directly beneath the AI processor. This ultra-short power path drastically reduces transmission losses, improves efficiency and supports the high current demands of next-generation AI processors—all without the need for costly infrastructure overhauls.
At the core of this innovation is AmberSemi’s miniaturized, high-efficiency power supplies, designed to fit within the limited space beneath the processor. This vertical integration not only eliminates nearly 50% of wasted power, but also enables more compact, scalable AI systems capable of meeting the extreme power densities required by future computer architectures.
Also Read: Confidential Computing vs Traditional Encryption: Key Differences Explained
[To share your insights with us as part of editorial or sponsored content, please write to psen@itechseries.com]
