Analysis
Look past the individual headlines from this week's funding roundup and a pattern emerges that's easy to miss story by story: the largest rounds aren't AI model companies at all. They're satellites, thermal batteries and financial infrastructure -- three sectors that have almost nothing in common on the surface but share the same underlying investor thesis, that the physical and financial bottlenecks standing between AI's compute buildout and its eventual return on capital are now where the biggest, safest checks are being written.
K2 Space's climb to a $6.8 billion valuation on the back of $1 billion in signed contracts is a satellite-manufacturing story on its face, but the demand driving it is increasingly defense and communications infrastructure that itself depends on the same data-processing capacity AI labs are racing to build. Antora Energy's $550 million round, meanwhile, is explicitly a power-infrastructure play -- the company's own materials frame its thermal batteries as a direct answer to AI data-center demand straining the grid, following the same logic behind Commonwealth Fusion Systems' climb to $4 billion in total capital raised and Joulent's $1.75 billion tie-up with National Grid Ventures earlier this year to power a Microsoft data center in Texas.
“Antora's round drew Liberty Mutual, an insurer, alongside its traditional cleantech backers.”
What's notable about this wave is who's writing the checks. Joulent's round came from a utility, not a venture fund. Antora's round drew Liberty Mutual, an insurer, alongside its traditional cleantech backers. These are institutional players whose core business depends on predictable, long-duration returns -- exactly the profile that has historically avoided venture-style bets on unproven physics or unbuilt factories, now underwriting them directly because the demand signal from AI infrastructure has become impossible to ignore.
CAIS's raise, though thematically distant from power and satellites, fits the same pattern one layer removed: it's financial infrastructure that lets a wider base of independent advisors access the private markets now funding all of the above. A $2 billion valuation on 37% organic growth is CAIS's own investors betting that more capital, not less, will keep flowing into exactly this kind of infrastructure-adjacent private investing over the next several years.
For allocators, the practical takeaway is that tracking "AI funding" purely through model-lab valuations increasingly misses where the actual capital concentration is happening. The infrastructure layer -- power, satellites, chips, and now the financial rails connecting private capital to all of it -- is where this week's biggest, most durable-looking rounds landed, and it's worth watching as its own category rather than a footnote to the headline AI labs.
