Former Intel CEO Pat Gelsinger made headlines at Ai4 2026 in Las Vegas with his blunt statement: “Today’s GPUs suck.” His remarks were part of a broader discussion regarding the limitations of current AI hardware, particularly focusing on the infrastructure that supports these technologies.
Gelsinger emphasized that the efficiency of GPUs is considerably lacking, claiming they are “very power inefficient” and “computationally limited.” He argued that the future of AI relies less on acquiring more accelerators and more on overcoming fundamental constraints in computing power, memory, energy, and economics.
During the keynote, moderated by New Yorker staff writer Gideon Lewis-Kraus, Gelsinger highlighted the crucial role of semiconductors in AI advancements. He stated, “There ain’t no compute without chips sitting underneath it,” reinforcing that chips are the essential fuel powering modern intelligence demands. He noted that the continued pace of AI development has been made possible due to decades of investment in foundational technology, including the internet, semiconductors, and cloud infrastructures.
Gelsinger outlined four primary priorities necessary for advancing AI infrastructure: improving efficiency, energy availability, growing semiconductor production, and enhancing economic frameworks. He criticized the industry’s current reliance on high-bandwidth memory (HBM), labeling it as inefficient but recognized it as the best available option for AI modeling at present.
Alongside Gelsinger, Sachin Katti, the head of compute at OpenAI, shared insights on the significant bottlenecks in infrastructure development. He conveyed the shift in focus from advocating for AI’s value to securing governmental and industrial buy-in for building the necessary support structures for AI technologies. Katti pointed out that the supply chain faces numerous challenges, from constructing data centers rapidly to ensuring adequate energy supply and memory capacity for chips, all critical components for AI deployment.
Energy was identified as a critical factor limiting AI’s progress. Gelsinger remarked that economic capacity is intrinsically linked to energy capacity in a digital economy. He asserted the need for increased energy efficiency to ensure AI computation is effectively powered, highlighting that nobody will invest in a data center without the assurance of sufficient energy supply.
Lastly, Gelsinger critiqued current economic conditions surrounding AI production, stating that the expenses associated with generating intelligence are excessively high. He argued that advancements in various domains like semiconductors, memory, and power systems need to converge to drive down costs dramatically.
In conclusion, Gelsinger’s message underscored that merely deploying more GPUs will not lead to the next phase of AI; instead, a robust infrastructure that allows for greater efficiency and capability in delivering intelligence is essential. The Ai4 conference continues this week, focusing on these crucial developments in AI technology and infrastructure.
Welcome to DediRock, your trusted partner in high-performance hosting solutions. At DediRock, we specialize in providing dedicated servers, VPS hosting, and cloud services tailored to meet the unique needs of businesses and individuals alike. Our mission is to deliver reliable, scalable, and secure hosting solutions that empower our clients to achieve their digital goals. With a commitment to exceptional customer support, cutting-edge technology, and robust infrastructure, DediRock stands out as a leader in the hosting industry. Join us and experience the difference that dedicated service and unwavering reliability can make for your online presence. Launch our website.