What happened
Databricks announced on July 12 that it will acquire training-infrastructure startup Lamini for $2.1 billion, consolidating enterprise fine-tuning and custom-model capabilities alongside its existing MosaicML acquisition. Lamini specializes in domain-specific model training for regulated industries with data residency requirements.
The deal is expected to close in Q3 2026. Databricks said the combined stack will offer end-to-end custom-model training on customer data without leaving the Databricks lakehouse.
Why it matters
Enterprise AI is consolidating around platforms that own the data layer and the training layer together. Another large acquisition signals that custom-model training is a strategic category, not a niche.
Regulated-industry focus also aligns with the sovereign-cloud and data-residency trends seen in Cohere's recent raise.
MintedBrain take
If you are already on Databricks, this may simplify custom-model workflows. If not, the acquisition is a signal about where enterprise AI spend is heading, not a reason to switch platforms on its own.
Discussion
Sign in to comment. Your account must be at least 1 day old.