Data Management
September 3, 2026
Sept. 3, 2026 — IBM and Confluent have announced the launch of IBM Granite Time Series models in Early Access on Confluent Cloud, bringing forecasting and anomaly detection directly to enterprise data streams.
The integration pairs IBM’s time-series foundation models with Confluent’s data streaming platform, so teams can analyze operational data as it’s generated. Teams can call the models from Apache Flink on Confluent Cloud, turning live signals into forecasts, anomaly alerts and downstream actions, all without standing up a separate machine learning environment.
Early Access starts on Confluent Cloud on AWS. Confluent Platform support for on-premises and hybrid environments is planned next.
Bringing Time-Series Intelligence to Streaming Data
Time-series data drives many everyday business decisions: forecasting demand, monitoring equipment, flagging suspicious transactions and tracking application performance.
Teams have traditionally had to build and maintain a separate model for each dataset or application. IBM Granite Time Series models take a different approach: a single set of models that works across varied time-series signals, supporting forecasting, anomaly detection, similarity search, classification, gap-filling and optimization.
Confluent supplies the live business context. Its platform continuously streams, connects, governs and processes real-time data, capturing live business signals that IBM Granite Time Series models use for anomaly detection and forecasting.
Key Benefits of IBM Granite Time Series Models
Running IBM Granite Time Series models within Confluent Cloud gives teams a simpler way to add time-series intelligence to streaming applications. There are five key benefits:
- Real-time intelligence where the data lives: Apply forecasting and anomaly detection directly to streaming data without first moving it to a separate machine learning platform or data warehouse.
- Zero configuration: Confluent manages model serving, infrastructure, scaling, and runtime operations – there is no provider credential to manage or glue between data pipelines and the model. Call IBM Granite Time Series models directly from Flink SQL for real-time anomaly detection and forecasting.
- Fresh, enriched context: Confluent continuously captures and processes data into an up-to-date view of the current state of the business – from sensor telemetry and payment activity to application metrics – so models can act on what’s happening now rather than stale batch data to make more reliable, accurate predictions. Inference results are written to Kafka topics and shared with fanout — consumable by alerting systems, dashboards, lakehouses and AI agents.
- Built-in governance and traceability: Inference pipelines adhere to the same schemas, lineage and access controls as everything else on the platform. Kafka topics are durable and replayable, which supports auditing, troubleshooting, model evaluation, and rerunning inference against historical data.
- Cost efficiency: Native inference can reduce the need for dedicated model-serving infrastructure or GPUs and avoid cloud ingress and egress fees.
A Portfolio for Different Time-Series Decisions
IBM and Confluent are bringing four complementary time-series foundation models to Early Access, as listed below. The portfolio is designed to support different needs across forecasting, anomaly detection, scale, accuracy and other time-series tasks.
Users can access supported models through Confluent’s existing AI_FORECAST and AI_DETECT_ANOMALIES Flink SQL functions and switch between models without redesigning the underlying pipeline.
The Early Access portfolio spans four complementary models:
- PatchTST-FM-r1: The all-around performer, strongest at probabilistic forecasting with full distributions and quantiles.
- FlowState-r1.1: Sampling-rate invariant, with the best point-forecast accuracy.
- TTM-r3: The best efficiency and performance trade-off — supports control variables and runs many series at low cost on CPU.
- TSPulse: Anomaly detection, classification, similarity search and gap-filling from a 1M-parameter model.
All four are small by design—1M to 260M parameters, GPU-free—which is what lets inference sit inside a streaming pipeline, and tuning on your own data or bringing your own weights works through the same interface.
Applying time-series intelligence to mission-critical enterprise decisions
- Forecasting and planning: A retailer managing thousands of products could use a shared time-series model to forecast demand across its catalogue, helping inform replenishment, allocation and pricing decisions without building a separate model for every product line.
- Anomaly detection: In financial services, transaction activity can be scored while a payment is still in flight. If unusual behavior is detected, the result can trigger the next action, such as blocking the payment, sending it for review or passing additional context to an AI agent. The same approach can be applied to application performance, network activity and equipment telemetry.
- Production optimization: Manufacturing environments generate continuous streams of signals such as temperature, speed, throughput and dosing rates. Time-series models can help teams forecast how changing conditions may affect production outcomes and use those forecasts to guide operating decisions.
- Semantic intelligence: Time-series analysis can also help answer a practical question: have we seen this pattern before? By comparing current activity with similar historical patterns, teams can identify previous production runs, demand shifts or confirmed fraud cases that resemble what is happening now. That context can then be used by people, applications or AI systems before the next action is taken.
IBM Granite Time Series models for forecasting and anomaly detection are available now in Early Access on Confluent Cloud. Organizations can apply the models to their own streaming data and work with IBM and Confluent as the capabilities continue to evolve.
Confluent, an IBM company, is the data streaming platform that is pioneering a fundamentally new category of data infrastructure that sets data in motion. Confluent’s cloud-native offering is the foundational platform for data in motion—designed to be the intelligent connective tissue enabling real-time data from multiple sources to constantly stream across an organization. With Confluent, organizations can meet the new business imperative of delivering rich, digital frontend customer experiences and transitioning to sophisticated, real-time, software-driven backend operations.
IBM is a leading provider of global hybrid cloud and AI, and consulting expertise. We help clients in more than 175 countries capitalize on insights from their data, streamline business processes, reduce costs, and gain a competitive edge in their industries. IBM’s breakthrough innovations in AI, quantum computing, industry-specific cloud solutions and consulting deliver open and flexible options to our clients. All of this is backed by IBM’s long-standing commitment to trust, transparency, responsibility, inclusivity, and service. Visit www.ibm.com for more information.
The Big Data Inside Amazon’s New Fire Phone
The new Fire phone that Amazon launched this week looks like your ordinary black smartphone,…
Three Reasons to be Scared of the Internet of Things
We know the Internet of Things forecasts: 50 billion connected devices by 2020. Apparently, there’s…
Wanted: Intelligent Middleware That Simplifies Big Data Analytics
We’ve seen tremendous technological innovation in the data analytics space over the past 10 years….
GPUs Tackle Massive Data of the Hive Mind
LIVE from GTC12 — The flock of birds that weaves seamlessly through the sky, propelled…
Why Hadoop on IBM Power
In the quest to achieve data-driven insight, Hadoop running on Intel X86-based processors has emerged…
Datanami’s Leverage Big Data Summit Wraps Up
Dialog and networking were on the Datanami agenda this week as we kicked off our…
AI Search Has a Benchmarking Problem. Qdrant Wants to Fix It at 10-Billion-Vector Scale
Vector database company Qdrant has released a dataset with 10 billion documents to help put…
‘Ghost Demand’ Prompts Texas to Halt New Data Center Grid Connections
Texas has put a hold on new data center grid connections as it tries to…
Data Centers Are Under Attack. Is the Anger Justified?
Anger over the construction of AI data centers has triggered political leaders to call for…
Cost per Token vs. Cost per Task: What Really Determines AI ROI
In 2018, when we were still living in the early age of artificial intelligence, technology…
AWS Has Plenty of Analytics Tools. So Why Buy DuckLabs?
AWS already has plenty of ways to analyze data. It can tap into Redshift, Athena,…
McKinsey Report: Enterprise AI Is Becoming a Two-Speed Race
The enterprise AI race is starting to produce some separation. While AI adoption continues to…
Current Stories
Current Stories
