Every month we feature members from the Everything Marketplaces community to help highlight their story, marketplace journey, and share more about what's ahead. In this member feature we highlight Ben Moore, who’s the Co-founder & CEO of GPU Trader. GPU Trader is a marketplace that helps organizations with GPU capacity connect with customers looking for high-performance compute.
‍

So what’s your background briefly and what led to the idea for starting GPU Trader?

I’ve been in tech for 25 years. I started coding and breaking computers as a kid and never really stopped. My background is in IoT and spatial computing, and I’ve had the chance to work across the spectrum: from scrappy startups to scaled operations at places like AWS and Apple. I’m a product guy, and I’ve always been drawn to complex problems, often very unsexy problems, and figuring out how to make these systems better for people.
‍
Matt, my co-founder, and I met at a fun, maybe even weird,  little startup in the Open Networking space. There, we were doing some work on building a hybrid multi-cloud network observability platform. Our time at this company ended, and we both went our separate ways, joining the 2023 fractional consulting trend. 
‍
Little did we know that three months later, we would join forces again at an early Infrastructure as a Service company and discover the problem that led to GPU Trader. 
‍
GPU Trader came out of a very real market gap, Matt and I saw. On one side, AI teams were scrambling for GPUs and paying these insane cloud markups or sitting on waitlists. On the other hand, there was this dirty little secret in the industry. There were datacenters and enterprises with idle or underutilized GPU capacity that could be monetized, but they didn’t want to commit that capacity to long-term contracts. We built GPU Trader to fix that. It’s a marketplace that connects unused capacity with real demand, with zero friction. No sales reps, no inflated pricing, ready to go.

What were some of the first steps you took to start GPU Trader? What were some of the biggest challenges that you faced?

We were fortunate initially because we were working with a company that had the problem that GPU Trader solves on the supply side. So our first step, while it sounds simple, was validating the problem. We went deep with people we knew in the space. AI startups, datacenters, researchers, and data scientists. We wanted to make sure the supply-demand mismatch we saw wasn’t just theoretical. And pretty quickly, it became obvious: everyone was frustrated with the current system.
‍
The biggest challenge we faced then, and now, is time. AI is moving so quickly. Intellectually, I knew it then, but at the start, I don’t think I gave it enough space in my mental model and anticipated how much of an impact it would have on us.
‍

What’s the problem that you’re solving for and how has building GPU Trader as a marketplace proven to be a great solution?

The core problem is fragmentation. On one side, you’ve got AI labs, researchers, and startups with surging demand for GPUs. These teams need compute now, not after a three-month procurement cycle with an army of sales bros. And they’re not looking to sign long-term commitments just to get started.
‍
On the other side, there’s a pool of GPU capacity that’s traditionally sold in 2–3 year contracts with hefty upfront payments to unlock reasonable pricing. That model doesn’t work for fast-moving teams who want flexibility.
‍
GPU Trader bridges that gap. We make it easy and low-risk for providers to list excess capacity, and just as easy for buyers to access it on demand, with clear pricing and no lock-in. Most importantly, we reduce time to value. You can spin up a fully controlled GPU instance on GPU Trader in under two minutes. That’s a big deal when the alternative is waiting 20–30 minutes just to boot a machine, or worse, running on a model-serving platform where you don’t control the environment at all.
‍

How have you approached solving the initial cold start problem and getting the flywheel spinning for GPU Trader?

I’m low to no bs. I don’t think we’ve cracked this one yet. It’s not lost on us that we chose a very difficult business model, and that GPU Trader has two sets of customers. One thing we got wrong was which side of the platform would be hardest to start… Honestly, we were convinced that it was the supply side. But that is not it.
‍
Back to my previous comment about the most significant challenge we face being time, the market has shifted so quickly since we founded the company that I don’t think anyone anticipated the prices and availability we would see for GPUs right now. Even with our most conservative modeling, we thought we had another 12-18 months until we saw what we see today. Competing for demand-side customers when everyone is offering GPUs at 20% of hyperscaler (AWS, Google, Azure) pricing is complicated. Especially when those hyperscalers are giving it away, it doesn’t matter how well priced you are when you compete with free.
‍
So we are still working on the cold start problem. We’ve shifted from enterprise adoption of AI to refine our GTM approach to rapidly build out a programmatic, bottom-up, go-to-market machine focused on developers, like the early days of cloud. Our tactics include: targeted outreach campaigns to directly engage developers in key AI communities, working with our trusted advisors specializing in developer growth, and integrating with technology partners to expand our reach and meet developers where they are today.
‍
It’s an awareness game, and this space is spicy. That said, when we got our first self-service customer at the end of Q1, that was incredible!
‍

What’s been your biggest learning so far?

Starting a company is unimaginably hard and rewarding, and I’m glad I waited until my forties to do it. I don’t think I would have had the maturity to take the time to take care of myself. While it feels like a sprint, this is a marathon, and carving out space to recharge is critical to my performance at work and home. You can take the boy out of Amazon, but you can’t take Amazon out of the boy, and I subscribe deeply to the philosophy of prioritizing long-term value over short-term gains.
‍

How big does GPU Trader get and what’s needed to get there?

This is a question Matt and I discuss regularly. Our vision is to become the trusted source for enterprise-class datacenter GPUs. How many cards do we connect to workloads every month to achieve that? I’m not sure we know yet. From a business perspective, we still have to decide how big we want to get.
‍
On the one hand, we see the potential to raise and build a $10B category leader. Isn’t that the dream? We know what it takes to get there, the investment, scaling, and commitment required, and we’d need full conviction before pursuing this route.
‍
The question becomes, is that the right path? Should it get that large? On the other hand, we also see a clear path to a $200M-$400M a year company. This path comes with a different set of capital requirements and expectations.
‍
The jury is still out. We are still evaluating which business we want to build, and more importantly, what version the market needs. 
‍

What’s exciting and ahead for GPU Trader that you can share with us?

In 2024, we focused entirely on the core platform. I am proud of what the team has built and of getting it to market as quickly as we did. 2025 is all about investing in our GTM and building awareness. We will continue to add to the platform, building by focusing on our customers and the direct feedback on their needs…lots of fun things coming soon!

‍
Connect with Ben in the Everything Marketplaces community. ‍‍