Data Center Solutions at Scale

Virginia has been an important place for Amazon and AWS going back to the establishment of Amazon Web Services in 2006, which is when our first data centers were established here. Since that time, we have invested $63.9 billion in Virginia from 2011 through the end of 2023, building out our AWS data center infrastructure there. That number, as we look at this year, is expected to be around $75.4 billion since 2011, so we’re continuing to invest heavily in the Commonwealth.

Welch: We thank you for that commitment to Virginia. To support that growth, you all have had an active partnership with Northern Virginia Community College (NVCC), creating an associate’s degree in cloud computing. Are there any other Virginia partners working with the company on workforce development?

Kellogg: That partnership with the Northern Virginia Community College system has really been a critical one for us. We first started our cloud computing associate degree partnership with them in 2018, when we established the first associate cloud computing degree in the nation. Really, that was just the beginning. That program has grown to over 13 community colleges across the Commonwealth that are now currently offering cloud computing programs to their constituencies. In 2019, we supported a collaboration between NVCC and George Mason University to create a pathway for community college students to earn a master’s degree. And that collaboration actually won the 2019 Partnership of the Year award from Higher Ed Dive, which was a nice early accomplishment and recognition of that program.

Another close partner has been school districts throughout the Commonwealth. We’ve worked with a number of institutions to offer AWS Girls Tech Days, to connect thousands of students around our data centers in the clusters that we’ve built with Women in Tech to give young girls a head start when pursuing a career in the technology industry. We’re super proud of that program. We’ve also established 38 learning labs called AWS Think Big Spaces, to give students in communities where our data centers operate a place beyond the classroom to get hands-on experience with technology.

Welch: How do data centers play a role in the proliferation of artificial intelligence?

Kellogg: Cloud is really foundational to artificial intelligence and the emergence of generative AI technologies. When you look at today’s AI and machine learning workloads, it requires enormous amounts of compute, storage, and networking capacity, all of which can be provided by large-scale data centers, of course, which AWS owns and operates. And AWS is known for its scale, and that’s why most companies that need access to on-demand infrastructure capacity at scale for AI workloads come to our company. 

Since we’ve been building large-scale data centers for more than 15 years and GPU-based servers for more than 12 years, we have a massive existing footprint for AI infrastructure around the world. Virginia is a very, very big part of that infrastructure.

Welch: What are the unique advantages that we have here that can position Virginia as a leading hub for AI innovation and development?

Kellogg: Historically, Virginia has positioned itself as an environment to make data centers flourish more broadly. It’s always been data center-friendly, and I hope that continues to be the case. It starts with focusing on the power of public-private partnerships. Government agencies are among the fastest-growing adopters of the cloud, and to better serve constituents and achieve their missions, they are leveraging technologies used by the private sector to improve that customer experience with citizens and constituents and of course the overall business operations of the Commonwealth.

Another key ingredient in Virginia’s competitiveness is the talent with the right technology skills to build on our progress. That’s a big part of why we’ve invested so heavily here and also have HQ2 here in Virginia. Virginia’s highly rated college and university system must continue to produce highly skilled, technical talent, and businesses like ours should be encouraged to do their part, like with some of the programs that I mentioned earlier. That’s why we have partnered with institutions and education at every level, ranging from the public primary and secondary schools to upskill programs created to help veterans develop in-demand cloud skills and other technical skills.

Welch: What best practices can companies follow to make sure they’re implementing AI in a responsible manner?

Kellogg: This is a really big priority for us. We, as a company, are committed to putting responsible AI into practice for research and continued education in this area. Responsible AI does require a multidisciplinary effort. At AWS, we take a people-centric approach that starts with education and building the next generation of developers and data scientists with programs like our AI and machine learning scholarship program, as well as new bias and fairness courses from our Amazon Machine Learning University.

In this process, we consider a number of factors for developing and implementing AI in a responsible manner. For one, it’s important to consider diverse perspectives, regulations, training, and education when building and using AI systems. Considerations that apply to all phases of the AI life cycle include promoting human oversight and control, which is very important, and cultivating a culture of responsible innovation. In the design phase of these technologies, it is important to define the use case and requirements for the AI system, establish performance criteria, and explore the potential impact of the system on users and other parties. 

In the building phase, it is important to develop the AI system by iteratively training and tuning, after careful evaluation, and to consider mitigation techniques to address potential risks. And having that risk-based approach is really, really important. And then finally, in the operation phase of leveraging and implementing these technologies, it is important to implement safeguarding mechanisms to continuously validate the system’s performance to use feedback mechanisms and consider human oversight.

Welch: Your experience in Virginia has taken you to many different regions and places throughout the Commonwealth. Do you have a favorite spot you’ve visited?

Kellogg: I have been fortunate to travel across the Commonwealth, so I don’t say that I have an absolute favorite. But one of my kids went to college in Williamsburg. I’ve always loved that area. The other went to Blacksburg, to Virginia Tech. But I love all counties in Virginia, and we’re now investing in a large number of them as we continue to grow our footprint across the Commonwealth, and looking forward to discovering more and more of Virginia as we do that.

Welch: Two very important universities in Virginia who are doing a lot in the AI space and developing that next-generation technology. I really appreciate the time today.

Kellogg: Thank you very much for having me. I appreciate it.

For the full interview, visit www.vedp.org/Podcasts