Databricks Acquires MosaicML, Thomson Reuters Buys Casetext

Hi friend! Got a few minutes? I‘ve got some fascinating AI and legal tech updates to share. Let‘s geek out over coffee ☕

Databricks Goes Big on AI – Snaps Up MosaicML for $1.3 Billion

In one of 2022‘s biggest AI acquisitions, Databricks has decided to splurge on MosaicML, acquiring the startup for a massive $1.3 billion.

As a data analyst myself, I see why Databricks is making this big bet. Here‘s a quick primer on what each company brings to the table:

Founded in 2013, Databricks offers a unified data analytics platform for building AI applications. Its open source engine Apache Spark powers data pipelines at 80% of Fortune 500 companies. Databricks reached a valuation of $38 billion in 2021.
MosaicML, launched in 2019, is laser-focused on deep learning. Its flagship product – the MosaicML Python Toolkit (MPT) – makes it easier to build, train and deploy neural network models.

In fact, over 3.3 million developers have downloaded MPT so far (based on Github data). This is an impressive reach in just a few years!

MPT in particular stands out for its ability to train huge models efficiently. For example, MosaicML‘s newest model MPT-30B has 30 billion parameters. According to benchmarks, it achieves state-of-the-art accuracy on natural language tasks using 5x less computing power than comparable models.

This technology nicely complements Databricks‘ capabilities in managing and analyzing large data sets. Integrating MPT into its platform gives Databricks new muscle for serving up powerful enterprise AI applications.

The competition Databricks faces likely prompted this acquisition too. Rival OpenAI is leading the AI race with chatbot ChatGPT and a valuation of $29 billion.

Buying MosaicML allows Databricks to juice its AI offerings with leading deep learning models and talent. The sticker price of $1.3 billion reflects the red-hot demand for AI currently.

In fact, AI startups attracted $93.5 billion in funding globally in 2021, up from $31.6 billion in 2020 according to research firm Prequin. Databricks is competing aggressively in this environment.

As an industry watcher, I expect we‘ll see more mega AI acquisitions as adoption grows. AI could generate over $13 trillion in business value by 2030 according to PwC analysis. With so much at stake, the AI land grab is on!

Thomson Reuters Bets on AI Legal Tech with Casetext Buy

Global information services heavyweight Thomson Reuters has also made waves, announcing the $650 million purchase of legal tech startup Casetext.

Let‘s examine what each party brings to the table:

Founded in 2004, Thomson Reuters provides news, data, and analytics to professionals in finance, legal, tax, and compliance. It made $6.3 billion in revenue in 2021.
Casetext, founded in 2013, develops AI-powered legal applications used by over 10,000 law firms and corporations. Its CARA Research Assistant leverages natural language processing to analyze contracts and regulations.

Legal AI is forecast to balloon from $1.7 billion currently to $5.8 billion by 2028 according to Precedence Research.

Thomson Reuters is positioning itself at the forefront of this seismic shift. Earlier in 2025, it partnered with Microsoft to launch Copilot, an AI plugin for professionals using Microsoft 365.

The Casetext acquisition builds on this momentum. CARA can rapidly scan and extract key clauses from complex documents. Just this year, Casetext unveiled CoCounsel – an AI lawyer based on GPT-3 that generates legal memos and briefs in seconds!

As an AI enthusiast, I find Casetext‘s technology seriously impressive. According to CEO Jake Heller, CoCounsel can perform work equivalent to a first-year associate at a law firm.

This level of automation frees up lawyers to focus on higher-value tasks and advice. My prediction – more law firms will embrace AI as the technology advances and payback period improves.

For now, Thomson Reuters is making big bets in legal AI both through acquisitions like Casetext and partnerships with AI leaders like Microsoft. With legal discovery often costing tens of thousands of dollars, AI automation makes a very compelling ROI case.

Pump Uses AI to Optimize Cloud Costs

Shifting gears, I want to highlight an emerging startup called Pump that caught my eye recently. Pump employs AI algorithms to analyze cloud usage and spending patterns to optimize costs, especially for AWS customers.

This data-driven approach really appeals to me as an analyst. Managing cloud costs is incredibly complex, with fluctuating spot instance prices and convoluted discount programs.

In fact, a Flexera 2021 survey found that enterprises underestimated their public cloud spend by an average of 23%! Getting unexpected giant cloud bills is the stuff of nightmares.

This is where Pump comes in. It acts like an automated cloud procurement manager, continuously identifying opportunities to utilize reserved instances, Savings Plans, and other discounts.

Pump claims it can save companies up to 60% on their AWS bill without performance impacts. So far it has helped over 100 businesses cut costs, including NFP and Blueboard.

The startup has raised $1.5 million from investors like Y Combinator. As more workloads move to the cloud, I expect AI-powered cost optimization will become critical for enterprises.

Gartner predicts public cloud spending will grow 20% in 2025 to total $494.7 billion globally. With stakes this high, AI cloud managers like Pump provide huge potential value.

The Key Takeaways on the AI Landscape

Stepping back, I see a few key themes across these stories:

AI adoption is accelerating across sectors beyond tech – from legal to cloud computing. The business case for AI automation is becoming stronger each year.
Leading enterprises are aggressively pursuing acquisitions and partnerships to boost their AI capabilities. With so much at stake, the AI land grab is on!
AI startups with innovative applications like MosaicML and Casetext are commanding huge valuations and buyout deals. Funding and exit values have skyrocketed in the AI sector.
AI still requires human guidance. Solutions like Pump use AI to augment humans, not replace them entirely. Hybrid human-AI approaches will be key going forward.

Well, those are my takes as a data analyst! Let me know if you want to geek out more about any of these topics. I could chat about AI all day. Till next time!