In this conversation, Anastassia and Rishab discuss the complexities and advancements in AI infrastructure.

Rishab explains the evolution of data centers, the impact of geography on their design, the importance of sustainability, and the critical role of security.

Rishab holds a B.Tech in Mechanical Engineering from IIT-Madras, India, and an MS in Industrial Engineering from Purdue University. Rishab has a proven track record of leading cross-functional teams to conceptualize, strategize, engineer, and execute large-scale, complex programs in data centers and fulfillment centers, managing projects with a total infrastructure value exceeding $100B.

Rishab is currently at Oracle, leading networking products for Project Stargate. Prior to this Rishab led AWS Data Center Infrastructure transformation into the AI era at Amazon. He implemented cutting-edge liquid cooling technology and pioneered sustainability-focused innovation across more than 400 data centers in over 25 countries, setting new industry benchmarks for efficiency and resilience. At AWS, he achieved a 12% reduction in large-scale events (LSE), significantly enhancing data center server availability and customer experience.

Since 2022, Rishab has been actively involved in the startup ecosystem.

Project Stargate is a $500 billion public-private venture led by Oracle, OpenAI, SoftBank, and MGX, aimed at building the world’s most advanced AI infrastructure through a nationwide network of large-scale data centers in the United States. Announced in January 2025, Stargate is designed to meet explosive AI compute demands by deploying 10 gigawatts of cutting-edge data center capacity, powered by millions of AI chips.

Terms and acronyms

Rishab uses several acronyms when talking about data centers. UPS stands for Uninterruptible Power Supply. Colo Providers are colocation providers. PUE is Power Utilization Effectiveness. GPU is a Graphics Processing Unit. WUE is Water Usage Effectiveness. CDU stands for Coolant Distribution Unit.

Takeaways

AI infrastructure is essentially a large-scale data center.

Data centers must maintain high reliability and security. Unlike a personal computer, which can be switched off, there is no pausing in data centers.

Data centers, as the backbone of AI infrastructure, connect to other key infrastructure components, including energy, water, and fiber.

Efficiency in data centers has significantly improved over the years.

Geographical factors significantly influence data center design, particularly in terms of cooling methods.

Building data centers involves complex negotiations with various stakeholders, including governments, communities, utility companies, telecom operators, and sustainability advocates.

Sustainability is becoming a crucial aspect of data center design. Specific metrics have been developed and implemented to enable the digital industry to closely monitor the construction and operation of data centers.

Physical and digital security are paramount in protecting data and privacy.

On-premises solutions remain relevant for certain businesses.

We must consider how AI and the evolution of edge computing necessitate a reevaluation of the concept of data centers.

Soft skills are essential for effective communication in engineering roles.

The demand for jobs in data centers is expected to rise significantly.

Chapters

03:59 Understanding AI Infrastructure

07:00 Evolution of Data Centers in the AI Era

10:01 Geographical Impact on Data Center Design

12:01 Building Data Centers: The Negotiation Process

15:34 Sustainability in Data Center Development

18:03 Security Challenges in AI Infrastructure

21:15 Balancing On-Premise and Cloud Solutions

24:03 The Future of Edge Computing

27:00 The Importance of Soft Skills in Tech

30:03 Closing Thoughts on AI Infrastructure

Information

@romyandroby

“Leading Through Disruption”

AI Edutainment

The AI Imperative Book

Romy & Roby Book