# Databricks' MosaicML Acquisition for $1.3B (2023)

In June 2023, Databricks acquired MosaicML for $1.3 billion to integrate its MPT models and training tools into its lakehouse platform, accelerating generative AI capabilities.

In June 2023, Databricks, a San Francisco-based data software company, acquired MosaicML, a startup specializing in efficient large language model training and deployment, for $1.3 billion. The acquisition aimed to integrate MosaicML's MPT (Mosaic Pretrained Transformers) models and training infrastructure into Databricks' lakehouse platform, enabling enterprises to build and deploy custom generative AI models on their own data. This move positioned Databricks as a significant competitor in the enterprise AI space, offering an alternative to proprietary models from [OpenAI](https://www.wikiprompt.org/wiki/openai) and [Anthropic](https://www.wikiprompt.org/wiki/anthropic).

The deal reflected a broader trend in 2023 where major tech companies invested heavily in generative AI startups. Databricks, founded in 2013 by the creators of [Apache Spark](https://www.wikiprompt.org/wiki/apache-spark), had already introduced its own open-source model, [Dolly](https://www.wikiprompt.org/wiki/dolly), in March 2023. The MosaicML acquisition provided Databricks with the technology and talent to scale its AI offerings, particularly for organizations seeking to train models on private data without relying on external APIs.

## Background and Strategic Rationale

Databricks had long focused on data analytics and the lakehouse architecture, which combines data warehouse and data lake features. By mid-2023, the rapid rise of [ChatGPT](https://www.wikiprompt.org/wiki/chatgpt) and other generative AI tools had created demand for enterprise solutions that could leverage proprietary data. MosaicML, founded in 2021 by [Ion Stoica](https://www.wikiprompt.org/wiki/ion-stoica) (also a Databricks co-founder) and others, offered a platform for training and deploying large language models cost-effectively. Its MPT models, such as MPT-7B and MPT-30B, were open-source and designed for commercial use, making them attractive to businesses.

The acquisition was announced on June 26, 2023, with Databricks CEO [Ali Ghodsi](https://www.wikiprompt.org/wiki/ali-ghodsi) emphasizing that the combined capabilities would allow customers to "own their AI" by training models on their own data. The $1.3 billion price tag was notable for a startup that had raised only about $64 million in funding, reflecting the strategic importance of generative AI.

## MosaicML's Technology and Team

MosaicML's core offerings included the MPT series of transformer-based language models, which were trained using efficient techniques like FlashAttention and optimized for inference. The company also provided a training platform that supported distributed computing on [AWS](https://www.wikiprompt.org/wiki/amazon-web-services), [Azure](https://www.wikiprompt.org/wiki/azure), and [Google Cloud](https://www.wikiprompt.org/wiki/google-cloud). The team, led by CEO Naveen Rao and CTO Hanlin Tang, brought deep expertise in AI infrastructure.

Post-acquisition, MosaicML's technology was integrated into Databricks' Data Intelligence Platform, announced in 2023. This platform combined lakehouse data management with generative AI capabilities, allowing users to fine-tune models on their own data using tools like MosaicML's training library.

## Impact on Databricks' Product Portfolio

The acquisition enabled Databricks to launch several AI-focused products. In 2023, Databricks introduced Dolly 2.0, an open-source instruction-tuned model, and later integrated MosaicML's models into its platform. By 2024, Databricks released [DBRX](https://www.wikiprompt.org/wiki/databricks-dbrx-launch), a mixture-of-experts model with 132 billion parameters, which leveraged MosaicML's training techniques. DBRX was designed to compete with models from [Google](https://www.wikiprompt.org/wiki/google) and OpenAI, offering open weights and strong performance on benchmarks.

Databricks also developed [Genie](https://www.wikiprompt.org/wiki/genie), a natural language interface for querying data, which used AI models to translate user questions into SQL. These products positioned Databricks as a provider of "data intelligence," where AI models are trained on and interact with enterprise data.

## Competitive Landscape and Market Response

The MosaicML acquisition was part of a wave of AI consolidation in 2023. Competitors like [Microsoft](https://www.wikiprompt.org/wiki/microsoft) invested heavily in OpenAI, while [Amazon](https://www.wikiprompt.org/wiki/amazon) and Google pursued their own AI strategies. Databricks differentiated itself by emphasizing open-source models and data privacy, appealing to enterprises wary of sending proprietary data to external AI providers.

The deal was well-received by investors, contributing to Databricks' subsequent funding rounds. In December 2024, Databricks raised $10 billion at a $62 billion valuation, and by December 2025, it reached a $134 billion valuation after a $4 billion Series L round. The company's revenue grew rapidly, reaching $6.9 billion annualized by June 2026, driven partly by AI product adoption.

## Integration and Subsequent Developments

Following the acquisition, MosaicML operated as a division within Databricks, with its tools becoming part of the Databricks platform. In 2024, Databricks acquired Tabular, a data management startup, and in 2025, it acquired Neon, a serverless database startup, further expanding its data infrastructure. The MosaicML technology remained foundational for Databricks' AI offerings, including the Agent Bricks suite for building AI agents, launched in 2025.

In 2025, Databricks entered partnerships with [Anthropic](https://www.wikiprompt.org/wiki/anthropic) and [OpenAI](https://www.wikiprompt.org/wiki/openai) to integrate their models into its platform, complementing its own open-source models. These moves reflected a strategy of offering multiple AI options to customers, while maintaining the ability to train custom models on-premises or in the cloud.

## Financial and Strategic Outcomes

The $1.3 billion acquisition was a strategic bet that paid off as generative AI became central to enterprise data platforms. Databricks' valuation increased from $43 billion in 2023 to $190 billion by August 2026, with a $7 billion revenue run rate. The company's focus on AI, enabled by MosaicML, was a key driver of this growth.

However, the acquisition also faced challenges, including competition from [AWS](https://www.wikiprompt.org/wiki/aws) and [Microsoft](https://www.wikiprompt.org/wiki/microsoft), which offered their own AI services. Databricks responded by deepening integrations with cloud providers, such as a 2026 partnership with Microsoft to use Azure and Cobalt processors.

## Legacy and Industry Influence

The MosaicML acquisition set a precedent for AI startup acquisitions, with other companies like [Nvidia](https://www.wikiprompt.org/wiki/nvidia-2023-acquisitions) and [OpenAI](https://www.wikiprompt.org/wiki/openai-cognition-acquisition) making similar deals. It also highlighted the importance of open-source models in the enterprise, challenging the dominance of closed models. Databricks' approach of combining data management with AI training became a model for other data platforms, influencing the development of [generative AI](https://www.wikiprompt.org/wiki/generative-ai) in business.

Today, MosaicML's technology is embedded in Databricks' products, and its team continues to contribute to AI research and development. The acquisition is remembered as a pivotal moment in the commercialization of large language models, demonstrating that enterprises could build and own their AI systems.

## External links

- [Wikipedia: MosaicML](https://en.wikipedia.org/wiki/MosaicML)

---
Source: https://www.wikiprompt.org/wiki/databricks-mosaicml-1-3b-deal
License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)
Last updated: 2026-10-29T15:58:32.78632+00:00