Baseten, a startup that provides application developers with access to Nvidia AI servers and model customization tools, is reportedly in discussions to raise $1 billion. According to The Information, the funding round would value the inference provider at $11 billion, including the new capital. This represents a potential doubling of the company's previous valuation, signaling continued investor appetite for the infrastructure required to run artificial intelligence models at scale.

The reported talks occur alongside a broader push to deploy high-performance computing in increasingly remote environments. In the aerospace sector, Aitech has announced upgrades to its space-grade supercomputer, while Starcloud is advancing a conceptual pathway to deploy a constellation of 88,000 computing satellites. Together, these developments—spanning terrestrial venture capital and orbital hardware—illustrate a growing imperative to distribute processing power closer to end users and data sources.

The premium on inference infrastructure

Baseten operates at a critical juncture in the artificial intelligence supply chain, bridging the gap between raw silicon and application deployment. By renting out servers powered by Nvidia, the dominant designer of the graphics processing units that underpin modern AI, Baseten allows developers to run and fine-tune models without managing the underlying hardware. The reported $11 billion valuation target underscores a shift in market focus from training massive foundational models to inference—the actual execution and querying of those models in live applications.

Securing a $1 billion capital injection would provide Baseten with the balance sheet necessary to secure highly constrained compute resources. In a market where access to advanced GPUs remains a primary bottleneck for software developers, infrastructure providers that can guarantee uptime and processing capacity command significant premiums. The scale of the rumored round suggests that venture investors view inference hosting not as a commoditized utility, but as a defensible layer of the AI ecosystem where early leaders can capture outsized margins.

Pushing processing power to the edge

While Baseten scales terrestrial data center capacity, parallel efforts in the aerospace sector highlight the extreme edges of the compute infrastructure buildout. Aitech, a manufacturer of ruggedized electronics for harsh environments, is upgrading its space supercomputer to handle more intensive data processing in orbit. Simultaneously, Starcloud's proposal for an 88,000-satellite computing constellation points to a long-term vision where data generated in space is processed in space, rather than being downlinked to Earth for analysis.

This orbital compute push shares a fundamental driver with the terrestrial AI boom: the need to reduce latency and manage massive data volumes. Just as application developers rely on Baseten to serve AI models quickly to terrestrial users, satellite operators increasingly require on-orbit processing to filter and analyze sensor data before transmission. Though Starcloud's constellation remains in the conceptual and early development phases, the dual trajectory of these industries indicates that the infrastructure layer of computing is expanding outward, driven by the specialized demands of both artificial intelligence and aerospace.

Whether through multi-billion-dollar valuations for terrestrial inference providers or the deployment of ruggedized servers in low Earth orbit, the architecture of global computing is undergoing a structural expansion. As the capital requirements for these infrastructure projects continue to scale, the market will test the viability of both Baseten's premium valuation and the ambitious orbital networks currently on the drawing board.

With reporting from The Information, SpaceNews, Payload.

Source · The Information