With his infrastructure company LT350, neurosurgeon and inventor Jeff Thramann is pursuing an unconventional approach to the data infrastructure market. Co-founded with former US Secretary of Defense Mark Esper, the company intends to erect modular mini data centres on the canopies of open-air car parks. Target locations include commercial properties such as shopping centres, hospitals and educational institutions, transforming previously unused airspace above parking bays into decentralised computing nodes.
Thramann, who holds more than 130 patents and previously built and sold medical technology firms such as U.S. Radiosurgery and Denver CyberKnife, engaged with photovoltaics and battery storage at an early stage. Observing power fluctuations in automotive battery systems, he developed patented storage cassettes that are integrated beneath canopies. Following ventures into solar energy and charging infrastructure, the surging demand for computing capacity led to the addition of graphics processing units (GPUs).
Modular computing capacity in the airspace above car parks
The construction model developed by LT350 entails assessing local grid connection capacities, manufacturing matching battery units, and constructing carports with rooftop photovoltaics. Enclosures are mounted beneath the support beams, housing GPU servers alongside battery storage. A typical canopy occupies a footprint of just around 2,000 square feet, whilst the servers can draw up to one megawatt of electrical power. Thramann explained that the batteries are charged during off-peak periods, and the installations operate off-grid during peak hours so as not to overburden the public power supply.
- —On-site data processing for applications requiring particularly low latency
- —No reduction in the actual usable parking bay capacity on the premises
- —Grid-friendly power buffering via integrated battery storage systems
Focus on data security and market entry
LT350 identifies healthcare as its primary market. Because sensitive patient data is subject to strict protection requirements during processing by AI models, housing computing hardware directly on site behind facility-specific firewalls offers high security standards. Alongside hospitals, Thramann lists financial services providers, public authorities and defence contractors among his target clients. Compared with large, centralised facilities in remote locations, he regards spatial proximity to end users as the essential advantage for delivering real-time AI inference.
In financial terms, LT350 remains in the pre-revenue phase and projects annual revenues of 12.4 million US dollars per 2,000-square-foot unit. A merger is planned with Auddia, an AI audio technology firm founded by Thramann that is publicly listed with a market capitalisation of 6 million US dollars. Going forward, LT350 is to operate under the umbrella of Thramann Holdings, which is valued at 250 million US dollars.
