I've done a lot of research and talks recently on the economics of AI - sometimes called "tokenomics". I'm also pretty keenly aware of the cost of AI at my own company. One of the major AI hot topics out there on the Silicon Valley interwebs is sovereign AI solutions based on open source or open weight models. The idea is that the cost of AI will increase significantly, and that the moat for AI solutions will be in the domain specific unique tuning of models.
One valid strategy is to rent data center space and colocate the actual bare metal in these shared spaces. But the more I think about it, that doesn't make sense for us so we intend to build out our own bare metal solution in our new Arlington, VA office that should be move-in ready July next year. I thought readers might be interested in how we positioned this for NVIDIA and AWS as we pitch them to support this build out.
We are a fly spec when it comes to these two companies, but we are very innovative and are also taking a unique view on AI when it comes to the intersection of human innovation and technological innovation. We are hoping this will convince AWS and NVIDIA to partner with us on this build out. If anyone has the contact info for Jensen Huang that would help expedite my pitch.
Sovereign AI for the Mortgage Industry, Kids, Young People, and Emerging Professionals
PhoenixTeam (www.phoenixoutcomes.com) intends to build a sovereign artificial intelligence (AI) solution to power its three year investment in technology, methodological, and human innovation. This paper presents the business case for our build out and answers questions posed by the Amazon Web Services (AWS) team at it seeks support from NVIDIA for our initiatives.
We specialize exclusively in the heavily regulated mortgage industry and have had this focus continuously since our three founders met in 2001. In 2015, the founders formed the company with the mission of freeing the American people from the bondage of joyless mortgage technology. We design, build, operate, modernize, and maintain the technology for all three federal housing agencies. We also have a successful commercial division supporting many of the largest and most innovative companies in the industry. In December 2023, we took a hard look at our business, saw that it was going to be eaten by AI, and pivoted the strategy of the company. The mission did not change.
The major barrier to mortgage innovation is and has always been compliance. With this hard strategy pivot, we explored, designed, and implemented a commercial product focused on managing mortgage compliance change. In our go-to-market (GTM) process, we discovered a major human barrier, one that we addressed with our growing AI education practice. To date we have educated about 3,000 mortgage professionals on AI and generative AI (genAI) in mortgage. This practice teaches established mortgage professionals to apply AI innovation and unlearn decades of old ways of thinking.
Along the way, we also uncovered the existential fear for the future of kids, young people, and emerging professionals. Instead of fearmongering and talk, we formed Human Futures (www.aifutures.com). The philanthropic arm of our company, the Human Futures initiative enables this generation to think creatively, work productively, and build in a world that has been fundamentally reshaped by genAI. Our first class of Human Futures interns (aged 17 to 23) graduated this summer, and our first group of kids (aged 10 to 14) completed our builder program in August.
In the past year, we have pushed the limits of agentic engineering with the goal of perfecting an autonomous software development lifecycle (SDLC). We discovered that the barrier to an autonomous SDLC is not traditional engineering, or even AI engineering, rather it is the process of conceiving of and iterating through an idea. We designed and implemented a unique value engineering process, replacing the traditional roles of product, testing, analysis, and design on an agile team. Our method puts the power of creation in the hands of business subject matter experts rather than holding it hostage in the mysterious agentic engineering process. Watch PhoenixTeam’s internal Value Engineer Bootcamp.
We anticipate that the cost of AI will increase significantly. We also need to be able to better control our spend and exert a finer grain of control over our solutions and methods. This has catalyzed our need to design and build out an on-premises, bare metal solution in our Arlington, Virginia headquarters. The build out of our expanded office space completes by July 1, 2027, and will include a dedicated server room designed to host an eight graphical processing unit (GPU) node that will support about 20 engineers. In addition, it will support a smaller, local learning lab for our Human Futures and mortgage AI education practices. We seek to partner with AWS and NVIDIA in this build out.