Sponsored by
What’s in today’s newsletter:
Snowflake’s AI Gateway auto-selects optimal models ⚖️
Databricks raises $5B, valued at $190B 💲
Dynatrace acquires Arize to boost AI observability 👁️
Also, check out the weekly Deep Dive - What is CDC (Change Data Capture)
AI agents built to get work done
Skydive agents take on work for you, using the same tools your team already uses.
Talk to your agent on the web, Slack, email, iMessage, or right from your terminal. Wherever you pick up the conversation, your agent keeps the context.
Put agents to work across customer support, sales, marketing, engineering, ops, and more.
Give your agent a job. They’ll take it from there.
SNOWFLAKE
TL;DR: Snowflake’s Cortex AI Gateway now autonomously selects and evaluates the best AI models for tasks, speeding deployment, reducing complexity, and enhancing accuracy to boost wider AI adoption across industries.
Snowflake’s Cortex AI Gateway now autonomously selects the best machine learning models for specific tasks.
The gateway continually evaluates model performance, removing the need for manual comparison by data scientists.
Automated model selection accelerates AI deployment and reduces operational complexity for organizations.
This enhancement improves AI accuracy and responsiveness, enabling broader adoption across various industries.
Why this matters: Snowflake’s autonomous model selection removes barriers for businesses lacking deep AI expertise, speeding up deployment and enhancing accuracy. This innovation not only simplifies AI integration but also drives more effective, data-driven decisions across industries, ultimately boosting competitiveness and operational efficiency in an increasingly AI-dependent landscape.
DATABRICKS
TL;DR: Databricks raised $5 billion, reaching a $190 billion valuation to expand generative AI, scale globally, and innovate, solidifying its position as a leading data and AI-driven tech startup.
Databricks raised $5 billion, achieving a $190 billion valuation, highlighting strong investor confidence.
Major investors like Franklin Templeton led funding, supporting expansion into generative and enterprise AI.
Funding will accelerate global scaling, product development, and workforce growth to meet rising enterprise demand.
Positioned as a top tech startup, Databricks aims to advance AI-driven insights across multiple industries.
Why this matters: Databricks’ $190 billion valuation signals immense investor trust in AI and data analytics as essential drivers of future tech innovation. This funding boosts its ability to scale, accelerate product development, and expand global reach, shaping AI integration across key industries and reinforcing its role as a cloud and AI leader.
OBSERVABILITY
TL;DR: Dynatrace acquired Arize to integrate AI observability into its platform, enhancing real-time AI monitoring, anomaly detection, and root cause analysis for improved reliability and transparency in AI-driven applications.
Dynatrace has acquired Arize to enhance AI observability by monitoring machine learning model behavior and detecting issues.
Integration combines Dynatrace’s infrastructure tools with Arize’s ML monitoring for comprehensive AI and software visibility.
The platform aims to improve real-time anomaly detection and root cause analysis for AI performance degradation.
This acquisition strengthens Dynatrace’s position in AI observability, addressing AI transparency and lifecycle management challenges.
Why this matters: Dynatrace’s acquisition of Arize significantly advances AI observability, enabling businesses to better monitor and manage AI model performance and risks. This integration enhances transparency, reliability, and trust in AI systems, responding to growing enterprise needs for comprehensive tools that oversee both software infrastructure and AI workflows.
Editors note: This story may not be within the realm of cloud databases in the strictest sense, but I think it is important as I am seeing less of a delineation between data and AI/ML in the enterprise.
EVERYTHING ELSE IN CLOUD DATABASES
Palantir Foundry vs. Snowflake: Data Battle
SingleStore Launches Aura AI for Affordable Data Tasks
Yotta joins Intellidb for sovereign AI databases
Elasticsearch Data Left Exposed, Risking Theft
Alteryx One debuts on Google Cloud Marketplace
Neo4j fuels India’s AI startup surge
Adastra earns Databricks BrickBuilder in key sectors
Databricks Boosts Data Governance with Bacancy Tips
ClickHouse Booms as AI Drives Massive Growth
Fabric Updates: Latest Features & Insights Unveiled
DuckDB v2.0 previews powerful server mode, triggers!
LakeBase adds hybrid vector-text search to PostgreSQL
Cyberhaven grows with 4 new execs to boost security
Talabat speeds insights with near-real-time cloud analytics
DEEP DIVE
A look into CDC and why it is important
“Gladstone, the Sydney server is slow”…
Such definitive words.
Most folks that have not interacted with databases at a dep level, have no idea about trace flags, execution plans, transaction isolation levels, clustered and non-clustered indexes, etc.
That was the world of the DBA where you are relied upon to keep database humming along.
What I am getting at is that the modern data practitioner has to conversant many more things now other than deep level database mechanics.
I am referring to things like data observability, understanding ETL/ELT, Python scripting/dataframes, and also the world of cloud computing. From the Azure cloud perspective, you have to understand Regions, resource groups, VNets, etc., when provisioning a database instance. Also one has to think about cost/FinOps.
The long preamble is to get to the notion of Change Data Capture.
What Is Change Data Capture?
Change Data Capture, usually shortened to CDC, is a method of identifying changes made to data in a source system and delivering those changes to another system as they happen, or close to real time.
The important word is change.
Instead of repeatedly copying an entire database table, CDC captures only the rows that have been inserted, updated, or deleted since the last capture point.
Imagine a customer table with 100 million rows. During the last five minutes, perhaps 500 customers changed their addresses, 200 new customers were created, and 20 records were deleted. Without CDC, a data pipeline might have to scan a significant portion of that 100-million-row table to determine what changed.
CDC attempts to capture just those 720 changes.
There are several ways this can be done. Some systems use timestamps or version columns to identify modified rows. Others use database triggers that record changes into separate tables. The approach generally preferred for larger production systems is log-based CDC.
Most transactional databases already maintain an internal transaction log for recovery, replication, and durability. SQL Server has its transaction log, PostgreSQL has the Write-Ahead Log (WAL), Oracle has redo logs, and MySQL has binary logs.
Log-based CDC reads those records and converts database activity into a stream of change events.
That stream might effectively say:
Customer 123 was created.
Customer 456 changed their address.
Order 789 changed from Pending to Shipped.
Customer 321 was deleted.
Those events can then be published through systems such as Kafka, Debezium, AWS DMS, Azure Data Factory, Oracle GoldenGate, or native cloud database services and consumed by downstream platforms.
This makes CDC especially useful for moving data from operational databases into warehouses, lakehouses, search systems, caches, and streaming platforms without constantly performing full data loads.
The architectural benefit is significant.
CDC reduces unnecessary data movement, lowers load on source databases, and allows downstream systems to remain much closer to the current state of operational data.
That last point is becoming particularly important for AI and agentic systems.
If an AI agent is making decisions based on enterprise data, the difference between data that is several hours old and data that reflects a transaction from thirty seconds ago can matter enormously.
In that sense, CDC is increasingly becoming the bridge between systems of record and systems that need to react to changes almost immediately.
Gladstone Benjamin
🚀 Work With Cloud Database Insider
Looking to reach CTOs, CIOs, and enterprise Data Engineers and Data Architects?
Limited sponsorship slots available each month.
