In a world increasingly driven by compute, where the cost of processing power can make or break entire businesses, one founder is building the financial infrastructure to stabilize the chaos. Carmen Li, founder and CEO of Silicon Data, is on a mission to bring clarity, structure, and transparency to the most volatile layer of modern AI infrastructure: GPUs.
Since its inception in April 2024, Silicon Data has quietly grown into one of the most intriguing startups in the compute ecosystem. The company recently raised $4.7M in seed funding from DRW, Jump Trading, Wintermute Ventures, and others, proof that the market is waking up to a problem Li has been observing for years.
From Bloomberg to the Bottleneck
Before founding Silicon Data, Carmen Li was deeply embedded in the data ecosystem at Bloomberg, leading global partnerships for data products. Her role put her at the center of financial institutions, cloud providers, and emerging AI startups, but it was a conversation with several AI founders that sparked the idea for Silicon Data.
“These startups were building cutting-edge models, but they were locked into SaaS pricing while their compute costs were all over the place,” Li recalls. “Their revenue was static, but their costs could triple overnight. There was no way to hedge.”
This was morethan a pricing mismatch. To Li, it was the sign of an immature market. A massive ecosystem growing without the financial scaffolding most industries take for granted.
“If American Airlines couldn’t hedge oil, they couldn’t price tickets,” she said. “Compute is going to be bigger than oil someday. But right now, it doesn’t have the basic financial tools that every other commodity does, no futures, no options, no visibility.”
So she set out to build them.
The Product Suite: Transparency as a Service
Silicon Data’s flagship product is Silicon Navigator, an institutional grade interface that tracks GPU rental prices, resale values, energy efficiency metrics, and daily price movements. It’s designed for decision-makers, from CTOs to hedge fund analysts, who need real-time visibility into GPU economics.
“If you’re deciding between renting GPUs or building your own data center, we help you model that decision with real pricing data,” Li explains. “We track market rates across regions, chip types, and configurations.”
A Data-Driven Foundation
Silicon Data doesn’t just scrape public pricing. It ingests, normalizes, and validates data from over 30 global sources. . Their platform combines machine learning for anomaly detection with human QA to ensure that pricing accurately reflects real market depth.
At the core of this process is T-Guard, a proprietary data pipeline system inspired by human T-cells. Just as T-cells detect and defend against threats in the body, T-Guard autonomously ingests, monitors, detects anomalies, and normalizes data across Silicon Data’s global infrastructure, ensuring data integrity and resilience.
“We’re agnostic to chip design,” Li says. “We just want data that’s deep enough and clean enough to be reliable. If we don’t trust the signal, we don’t publish.”
Everything is modular, from the free dashboard to a paid API tier and enterprise pricing models. Additionally, the company has grown entirely through inbound interest. “We’ve had hedge funds, chipmakers, and data center operators reach out to us organically,” she says. “Some of our most-used tools were built from their requests.”
A Vision Bigger Than Benchmarks
What Silicon Data is really building is the financial infrastructure of compute, a new layer of pricing intelligence, benchmarking, and risk mitigation for an industry that’s still flying blind.
“If compute is the new oil, the tools to trade, benchmark, and hedge it should already exist,” Li argues. “Our vision is to make those tools accessible, just like they are in energy, agriculture, or metals.”
Li sees a near future where compute contracts are traded, benchmarked, and risk-managed just like oil futures or corn options. And she wants Silicon Data to be the neutral source that enables that transformation.
“We’re not here to speculate. We’re here to build the rails that make this market fair, efficient, and transparent.”
