Gerdau reduces data processing time from 1.5 hours to 12 minutes and cuts data costs 40% with Databricks
The problem
Gerdau's homegrown and open-source data tools were complex, disconnected, and hard to manage, requiring Python and Spark proficiency that limited adoption across the business. The platform lacked real-time processing for digital twin projects, and siloed teams created duplicate databases, driving data inconsistencies and a rising total cost of ownership.
First attempt
Gerdau's proprietary and open-source tool ecosystem was fragmented and overly complex, requiring specialist skills that blocked broad adoption, and its lack of real-time processing capabilities prevented the digital twins use case from being realised.
Workflow diagram · grounded in source
1
Delta Lake data foundation
Integration
Delta Lake sets the foundation for Gerdau's new unified underlying data infrastructure.
▾ source quote
“Delta Lake set the foundation for Gerdau's new underlying data infrastructure”
2
Delta Sharing data exchange
Integration
Delta Sharing enables Gerdau to easily and securely share data internally and externally with partners.
▾ source quote
“the company leverages Delta Sharing to easily and securely share data internally and externally with partners”
3
Photon fast query processing
Integration
Photon, the next-generation query engine, reduces average data processing time from 1.5 hours to 12 minutes.
▾ source quote
“Gerdau has reduced their average data processing time from 1.5 hours to 12 minutes — a huge performance gain and cost savings, as certain tables in their workflows are processed daily”
4
Unity Catalog governance layer
Validation
Unity Catalog establishes data governance standards with fine-grained access controls, data lineage controls, and access segregations across manufacturing processes.
▾ source quote
“With Unity Catalog, we have established data governance standards across our manufacturing processes. We have also implemented fine-grained access controls, data lineage controls and access segregations for different groups of users”
5
Power BI self-service reporting
Output
Power BI integration enables Gerdau's business teams to more easily access data and create their own reports and dashboards.
▾ source quote
“Unity Catalog has paired well with their integration with Power BI, further enabling Gerdau's business teams to more easily access the data they need to create their own reports and dashboards”
6
ML and generative AI workloads
Ai action
Databricks enables advanced ML applications including predictive maintenance, image and text classification, and solutions powered by generative AI.
▾ source quote
“Leveraging Databricks for advanced analytics and machine learning has enabled Gerdau to further explore cutting-edge applications beyond digital twins, such as predictive maintenance, image and text classification, and other solutions powered by generative AI”
7
LLM upskilling assistant
Output
One of Gerdau's first achievements using large language models (LLMs) is an assistant to help people on their journey for re/upskilling.
▾ source quote
“one of their first achievements using large language models (LLMs) is an assistant to help people on their journey for re/upskilling”
Reported outcome
Databricks delivered a 40% cost reduction in data processing and 80% in new streaming solution developments, cut average data processing time from 1.5 hours to 12 minutes, reduced ML model creation effort by 30%, and onboarded over 300 new global data users.
Reported metrics
Data processing cost reduction40%
Streaming solutions new development cost reduction80%
Average data processing time (before)1.5 hours
Average data processing time (after)12 minutes
Show all 6 reported metrics
data processing cost reduction40%
streaming solutions new development cost reduction80%
average data processing time (before)1.5 hours
average data processing time (after)12 minutes
ML model creation effort reduction30%
new global data users onboardedover 300
Reported stack
DatabricksDelta LakeDelta SharingPhotonUnity CatalogPower BI
◆ Does this fit your context?
Compare to your context
Tell us your scale, team, and constraints. We'll show what changes at your size, what fails at your scale, and whether this case is a fit, needs adaptation, or won't scale to you. Free demo, no signup.
Databricks delivered a 40% cost reduction in data processing and 80% in new streaming solution developments, cut average data processing time from 1.5 hours to 12 minutes, reduced ML model creation effort by 30%, and…
What tools did this team use?
Databricks, Delta Lake, Delta Sharing, Photon, Unity Catalog, Power BI.
What results were reported?
Data processing cost reduction: 40%; Streaming solutions new development cost reduction: 80%; Average data processing time (before): 1.5 hours; Average data processing time (after): 12 minutes (source-reported, not independently verified).
What failed first in this deployment?
Gerdau's proprietary and open-source tool ecosystem was fragmented and overly complex, requiring specialist skills that blocked broad adoption, and its lack of real-time processing capabilities prevented the digital t…
How is this supply chain AI workflow structured?
Delta Lake data foundation → Delta Sharing data exchange → Photon fast query processing → Unity Catalog governance layer → Power BI self-service reporting → ML and generative AI workloads → LLM upskilling assistant.
This case is one data point. Whether its pattern fits you depends on your volumes, your stack, and your exception load — that comparison is the step no case study can do for you.