This month, we cover ElasticON takeaways, Databricks’ $10B war chest, MongoDB’s AI grab, IBM’s DataStax buyout, and Turso losing a production database (oops). Plus, ClickHouse’s 30% price hike and why pre-LLM AIOps anomaly detection was all hype, no help. Buckle up.

Inside this issue

• Community updates

• Interesting reads

• Updates from the database industry 

Community updates

• Last month, we dispatched our GTM Lead, Antoni Olendzki, into the heart of ElasticON London—an event that, for those attuned to the shifting sands of data infrastructure, felt less like a conference and more like a live-action systems upgrade. The venue buzzed with 500+ engineers, architects, and tech execs, all converging around a singular, if unspoken, realization: the old ways of wrangling data are starting to buckle under modern demands. As always, migration and security loomed large—two eternal constants in the life of anyone dealing with enterprise-scale databases. But this year’s emergent phenomenon? RAG—Retrieval Augmented Generation. Like all good acronyms, RAG promises much: AI-enhanced search, smarter retrieval, the dream of a system that actually knows what you meant instead of just throwing regex at the problem. It was the talk of the halls, an undercurrent in discussions on Elastic monitoring and custom integrations. For Quesma, it was an opportunity to listen, question, and plot the next moves in the ongoing struggle against observability chaos. If you’ve got thoughts on where all this is headed—or just need a sparring partner for your next infrastructure overhaul—let’s talk. 

• Our founding engineer, Przemysław “Przemek” Delewski, hit the stage at Warsaw Gophers Meetup to break down Go Compile-Time Auto-Instrumentation—a better way to add observability without touching your code. The problem? Manual tracing is a pain, and runtime auto-instrumentation is tough to implement (both add overhead). The solution? Inject telemetry at compile time. No runtime hacks, no brittle patches—just built-in observability from the start. Przemek covered the evolution from early Instrgen failures to the new toolexec approach, which modifies Go code as it compiles. Alibaba and Datadog are in—this is shaping up to be a big deal. 

• We released Quesma 1.1.3 🎉

Interesting reads

• Our CEO, Jacek Migdal, wrote a deep dive on ClickHouse’s January 2025 pricing update—a ~30% hike plus new egress fees. Turns out, this hit a nerve. The post shot up to #2 on Google, got shared by Altinity, Inc., Edge Delta, and more, and is still sitting in the top 5. Why the buzz? Database pricing matters. ClickHouse Cloud is getting more expensive, and users need to know their options. Jacek’s post breaks it down—who wins, who loses, and how to keep costs in check.

• Vendors have spent over a decade promising that AI will revolutionize observability—automatically catching issues before they escalate. Spoiler: it didn’t happen. In the first part of our AI in Observability series, Jacek Migdal breaks down why AIOps and anomaly detection have mostly flopped, why P99 latency still reigns supreme, and which AI-powered features actually proved useful. If you’ve ever been burned by an “AI-powered” alerting tool—or just want to know what’s next—check it out.

Updates from the database industry 

• MongoDB just bought Voyage AI in a $220M deal, bringing advanced vector search in-house to power AI-driven applications. The move boosts Atlas’ ability to handle unstructured data and improves semantic search, making it a serious competitor in the AI database race. Voyage AI’s embedding models reportedly outperform OpenAI and Cohere by 10–20%—a strong signal that MongoDB is going all in on AI-powered search and retrieval. With Snowflake and Databricks as early investors, this acquisition was always going to be strategic. AI isn’t just about models—it’s about fast, accurate data retrieval. MongoDB wants to be the database for AI workloads. Now the question is: Can they pull it off?

• Turso’s AWS-backed service lost customer data, and the internet is not amused. The issue? No backups—a fundamental failure for any database, beta or not. Reactions range from rage (“A DB company has ONE job!”) to memes (“Was Redundancy Theory purely theoretical?”), but the core lesson is clear: never trust a beta DB in production—especially without your own backups. For indie devs questioning if they really need a trendy SaaS DB instead of self-hosted Postgres, this was an eye-opener. EdgeDB? PlanetScale? Turso? Or just plain Postgres? The debate rages on.

• LaunchDarkly just acquired Houseware (undisclosed amount) and announced a Snowflake integration, signaling a big bet on warehouse-native product analytics. The goal? Bringing feature flags and experimentation directly into the data warehouse. Houseware (winner of Snowflake’s 2022 startup challenge) specializes in no-code, AI-powered analytics, making it easier for product and engineering teams to run experiments on live data and track real business impact. With this move, LaunchDarkly isn’t just about feature toggles anymore—it’s expanding into product intelligence, AI-driven insights, and deeper data integration. Feature management meets real-time analytics.

• IBM is acquiring DataStax, adding Apache Cassandra-powered NoSQL and vector search to its watsonx AI stack. The official pitch? Helping enterprises unlock unstructured data for generative AI. But the real gem might be Langflow—DataStax’s open-source low-code AI app builder that makes it easier to build RAG and multi-agent AI applications (fun fact: DataStax acquired Langflow less than a year ago). With Cassandra, vector search, and Langflow, IBM is making a play for AI-native databases and app development. Whether this is a strategic masterstroke or just another enterprise AI grab remains to be seen.

• Databricks raised a massive $10B Series J, valuing the company at $62B—and Meta is in (it happened in late Dec 2024, but in case you missed it…). The deal deepens Databricks’ ties to Meta’s Llama AI models and underscores just how critical data platforms are to AI’s future. With $3B+ in revenue, 60%+ YoY growth, and non-dilutive financing, this isn’t just another funding round—it’s a power move. Databricks is stacking cash, likely gearing up for acquisitions, AI bets, and a 2025 IPO. The endgame? Owning the entire AI data stack—from storage to compute to model training. If you’re tracking the AI infrastructure war, keep an eye on this one.