We’ve got Databricks, a San Francisco-based data storage and management startup, laying down a cool $1.3 billion to snatch up MosaicML, a budding star in the generative artificial intelligence space. They’re placing their bets on the burgeoning demand for businesses to craft their own ChatGPT-esque tools – a wise move if I do say so myself.
The Deal of the Century
Databricks and MosaicML are essentially combining their powers like some kind of AI superhero duo. Databricks brings to the table its AI-friendly data management tech, while MosaicML adds its language-model platform into the mix. The endgame? To empower businesses to build budget-friendly language models of their own, using their very own proprietary data. Currently, the norm is for businesses to rely on third-party language models, trained on vast amounts of publicly available data found online.
MosaicML, which has been around since 2021, is set to become a standalone service under the Databricks banner. The focus of this startup has been to slash the cost of utilizing generative AI from an eye-watering tens of millions of dollars to a slightly less gut-wrenching hundreds of thousands per model. Not a bad aim, if you ask me.
The Rise of Generative AI
Let’s talk about generative AI applications. These are designed to churn out original text, images, and computer code based on users’ natural language prompts. Since the launch of OpenAI’s ChatGPT, a generative AI chatbot, in November, interest in this tech has exploded.
Today, businesses are building generative AI apps on top of ready-made language models licensed to them by companies like Anthropic and OpenAI. The commercial demand for these models is through the roof, opening doors for startups like MosaicML that claim they can offer similar models but at a lower cost and with customization to a company’s specific data.
The Power of Data
Data. It’s always been the key to success, and the need for it has only increased with the advent of large language models. Corporate tech leaders are under pressure to get their data in order for AI models, as data serves as the foundation for all algorithms.
Companies are already using Databricks for their data pipeline and feeding that information to MosaicML to train a code generation model. Databricks’s technology, known as lakehouse, prepares and manages business data for AI applications and unifies data, analytics, and AI programming tools in one system.
The Future of Generative AI
The global generative AI market is predicted to hit $42.6 billion by year-end, with a projected compound annual growth rate of 32% to reach a whopping $98.1 billion by 2026. That’s some serious growth, folks. Venture funding in generative AI startups has already shot up from $4.8 billion in 2022 to $12.7 billion in the first five months of 2023 alone.
Companies are increasingly finding value in “domain-specific” models, as they contain more industry terminology and know-how. There’s an expectation that businesses can spend significantly less by using smaller, pre-trained models, rather than building on the vast corpus of data that large language models like OpenAI’s ChatGPT offer.
Conclusion
So, what’s the takeaway here? AI is evolving, and fast. The demand for generative AI models is skyrocketing, and companies like Databricks and MosaicML are stepping up to meet this demand. It’s a brave new world out there in AI land, and these businesses are showing they’re not afraid to take the lead.
