We are at an inflection point in the revolution of how we as the service providers are operating today. For the last 35 years we have operated in the internet age and the internet revolution. The Internet revolution has shaped the likes of me and my generation, yes 80’s babies, we got to bridge the analog world and the digital world and for IT career professionals like me we built the internet age. We created this behemoth, unlocking infinite potential for eCommerce to healthcare and then Jobs delivered it to the palm of our hand and further accelerating the growth. In my adult age we went from only having pizza and Chinese food delivered to your doorstep by teenagers to double dashing (at no additional cost) 2 different restaurants because my kids want chicken nuggets from Mcdonalds versus what I ordered.
The last 22 years, we have done the same thing, just bigger, better, badder. Moore’s law got blown away by NVIDIA by fundamentally coupling the GPU to the CPU. Novel idea, right? Let the CPU do the instructions for the math for parallel processing and GPU do the parallel processing. From thin air it created an industry that according to UN Trade and Development commission will hit $4.8 Trillion by 2033. In reference, the entire GDP of Germany is 5 Trillion…that’s 3rd on the list after US and China.
Amazing thing about this number is that we didn’t know what iPhones are going to do to the internet age, we have no idea what Ai applications are going to do this Ai revolution. We are at the infancy of Ai, it’s a concept stage. We are still unlocking the thought of the potential that Ai must change our lives. Remember when I mentioned Moore’s law? We must come up with a new law that makes sense of how the Ai age follows before Nevens law takes over (oh yes, like it or not Quantum is around the corner).
Current state:
Currently we are working on creating larger models that can continue to enhance the number of neural networks. Companies like Open Ai / Anthropic / Gemini / Deepseek / SkildAi are all focused on scraping the internet to train their models, specifically to create MOEs (Mixture of Experts). MOE models are general purpose models that your ChatGPTs and Claude are built upon, allowing you to create images of your cat flying through space in a space suit. Space cats are steppingstone to the real purpose of why these models are getting such a large investment, AGI. Our pursuit of AGI goes back to why humans were the species that infected our world over all others, our incessant need to pass along our DNA to the next generation and ensure its survival. Scraping is a great method to get started on mimic’ng AGI, however we expect these companies in the next 2 to 3 years to achieve a degree of AGI.
Super Soldiers:
Every large enterprise is investing in figuring out how to get repetitive tasks to done by Ai Agents for their own workflow. There is a massive pressure on CTOs to deliver a working strategy for Ai to augment or increase productivity by Ai. A worker armed with the current Ai tools is more productive than one without and that is where we see as the true IT delivery for most enterprises headed.
Let’s take an a few examples, I’ll start with our own internal software development team. At the beginning of 2025 I had budgeted to hire at least 2 to 3 software development headcount, active coders. Then I discovered Cursor, rolled it out to our existing dev team and increased code development tremendously. Our software developers were rolling out bug fixes in minutes instead of 2-week iterations, overnight out-modding (reference from the robot’s movie ---yes, I watched it on repeat because my girls loved it) Agile development and birthed vibe coding! All of a sudden we didn’t need any more headcount for organic growth of our dev team, replacing a budgeted 400K per year headcount line item by $39.99 per month per developer. If I don’t account for how much more code we were able to ship (intangible asset value), just the tangible savings to STN is 8000%!! STN development team only consists of 5 active coders and a manager right now, imagine what organizations like Microsoft and Google are going to save.
Cursor by the way is the tip of the iceberg, it’s a beta Ai application in this lifecycle of Ai tools. Every day its getting better as the underline models get better, I played around with the new Gemini 3.0 Antigravity (outside of a better reasoning LLM) and it surpasses Cursor by a mile. It incorporates all the processes like documentation to code review to security reviews that were manual for Cursor armed software developers. The rate of change here from the advent of cursor to antigravity is 9 months!!!! Side note, this is not a product review document, this is a slice in time snapshot of these two tools, I expect Cursor to throw down a bigger blow back to Gemini in the next release and iteration….I love capitalism!
3 Letter agencies:
While the large models continue to refine, there is an influx of organizations that have sprung up in the ‘agents’ space. Ai agents play pivotal role in achieving AGI. If LLMs are the brain, agents are the arms and the legs. Currently, the right arm doesn’t know what the left is doing but soon the nervous system will complete, and messages will be coordinated and deliberate and will be able to send the right signals to the agents. And the agents will work in cohort to allow LLMs to interact in the digital world and above all interact with each other. MOEs are great for the replacement of what we currently use google search for, however when you start focusing on solving the real-world problems and use cases like lifesaving drugs development, food supply shortages, reversing climate change, energy shortage, you need more focused LLMs. Above all you need those LLMs to be able to communicate with each other and be able to interact with the dataset that they are training upon.
That’s where I believe agents will play a pivotal role, we as humans are creating data at an immense pace and for any LLM to train properly, the first step is to feed it tagged data. That’s why you saw Zuch buy up a 49% stake in Scale Ai for 14 billion dollars and made Alex Wang the head of their Ai Lab. Like your kid, your LLM is only as good as how good their education system is and with any luck, they will grow up to solve world hunger or stab you. Getting back to the point, agents will continue to get better in the digital space, and they will interact and reason and solve operational issues to streamline development of different work products and they all require…..you guessed it, GPUs.
Terminator
Meanwhile Physical Ai is upon us, time to build the Terminator. Robotics has been around for half a century, and they are doing some amazing things from microscopic surgeries to manufacturing assembly-line automation. Majority of the robotics today are used for industrial use cases and are pre-programmed to follow a set of instructions and that’s about it. An arm that designed to pick up an object from one bin to another can only perform that task. It does it perfectly, with a 100% consistency and efficiency. However, if the bin shifts slightly or the dimension of the object changes the arm can’t adjust its parameters and adopt. That’s where human intervention comes in and requires humans to interact with the process and the robot.
This is where we have partnered with Skild Ai, who is founded by two CMU professors to design a LLM from ground up. They are building a brain for general purpose robotics to give the ability for robots to handle things that we take for granted. Like, how we adopt if we miss a step going down the stairs. Might seem simple, but we don’t ever think about it. Our internal haptic system immediately reacts and tries to grab the railing and shifts our leg to get a stable footing. I have seen these developments firsthand; this is what led me to come up with an analogy that was featured in a video created for HPE Discover (you can view it here). With the fundamentals of how we as babies learn to walk, run, fear things that are bad are now being incorporated into the Skild brain allowing robots to be a vessel of the Physical Ai Agent. Like the digital Ai agents I talked about earlier, companies like Skild are building the physical Ai to cross the threshold of AGI from the digital space to the physical space. Now, immediately your thought went to Terminator or Matrix. The machines are NOT going to take over the world; they have no reason to even though Hollywood might want you to believe so.
What’s next?
If you are still with me, let me tie it all together. They’re our 3 major parallel efforts occurring right now; Ai tooling that is creating a new generation of engineering and accelerating R&D, time to market in every technology vertical, Ai agents (compounded with the Ai tooling) that can automate every workflow imaginable and last but the most important is Physical Ai. Within the next 5 to 10 years, we project all 3 verticals converging into a singularity – Superintelligence transcending from digital world to our 3D world.
Everything that I have laid out so far requires computational power unlike what we have ever seen. Best way to explain it is the cost of the power grid at the turn of the last century. We don’t have a good idea of what it costs to build our power grid, but we do have an idea to re-haul and modernize our power grid (US only) will range anywhere from $2 Trillion to $5 Trillion. So, when you look at the spend that the market is up in arms regarding Open Ai’s $1.5 Trillion spend commit, it’s a drop in the bucket. We are just at the point of building the first few power plants and the fight to figure out if Edison’s DC current should be the standard or Tesla’s AC let alone every invention that uses electricity hasn’t even been thought off.
Where does STN play in all of this?
STN is uniquely positioned at this juncture as one of the first utility companies. Our goal is to own and deliver distributed computational power to Ai labs and in parallel build an extensive network across the world to deliver superintelligence and robotics Ai cloud. In conjunction we want to create autonomous data centers including robotics manufacturing. Everything that I talk about above requires GPUs and a comprehensive network to operate it across the world, as a NVIDIA NCP preferred partner and our partnership with Skild Ai gives us a leg up to achieve our goals.
Here are the verticals STN currently operates in, and we are looking for growth capital to grow the business rapidly:
STN is focused on GPU-as-a-service, distributed data centers (expanding from 4 sites), autonomous data centers, and US-based robotics manufacturing. We are leveraging NVIDIA’s Cloud Partner (NCP) program – where preferred partners like STN get priority access to AI infrastructure blueprints for scalable GPU clouds. Partnering with Skild AI gives us an edge in physical Ai. Our key differentiator from any of the hyperscalers like AWS or Microsoft is the enablement of up-and-coming Ai labs where AWS is focused on their other business units like Amazon. STN anticipates rapid revenue growth by 2030, delivering strong ROI for investors committed to this transformative sector. We are bullish on our strategy to slingshot STN from a regional NVIDIA NCP to a global power house delivering Ai utility to all corners of the universe (sneak peek to 2030’s plan ☺).
