Sociedad Química y Minera de Chile deploys pipelines 99% faster, scaling real-time data to new markets

“Fivetran gives us confidence in our data pipelines, so we can stop worrying about outages and focus on building AI models and expanding globally.”
— Jorge Utreras, Head of Projects, Architecture and Data Office at SQM

Sociedad Química y Minera de Chile (SQM), is a global mining leader and Fortune 2000 company headquartered in Chile, operating across 5 business lines: from lithium and iodine to specialty plant nutrition and solar salts. With more than 7,000 employees and expanding operations into new global markets, SQM is a critical player in the technologies driving the world’s transition to clean energy and sustainable development.

Supporting that transformation is Jorge Utreras, Head of Architecture and Data, who is responsible for enabling data-driven decisions across SQM’s global operations. His mission is to ensure that timely, trustworthy data reaches internal stakeholders to support SQM’s lean management culture and real-time decision making.

As SQM expanded globally, the complexity of its operations made it difficult to deliver consistent, real-time insights across remote mining sites and regional offices. These sites provide the data that powers production, logistics, and sales forecasting — making data delays a direct risk to operational efficiency and revenue growth.

Replacing fragile pipelines with a modern data stack

SQM’s original architecture started as an organic portfolio of solutions custom-designed for each line of business. Some groups used Google Data Stream to sync Oracle data into BigQuery, while others centralized workloads in Microsoft Azure. Despite serving each team’s particular needs, costs were high, performance was inconsistent, and every pipeline was a custom development project. Connectivity issues at remote mining sites frequently broke pipelines and recovery required, on average, 2 days of manual work. Weekend outages often stalled production and logistics until the following Monday, delaying decisions, deliveries, and customer responsiveness in a business that runs 24/7.

Initial patience with delays gave way as stakeholders came to realize just how critical real-time data was to making better, faster decisions. Utreras kicked off internal pilot programs demonstrating the possibility of fresher, more timely data, leading business teams to demand near real-time access for leaner, faster decision making. Utreras turned a technical problem into a business-critical opportunity.

He and his team needed a data movement solution they could trust, access, and manage easily. They partnered with Databricks, who recommended Fivetran to help modernize SQM’s data pipelines. They set up a pilot across 3 high-value data sources to test Fivetran’s ability to handle the company’s most challenging edge case: inconsistent network conditions at remote mine sites.

Within 30 days, Fivetran’s idempotent pipelines proved resilient, delivering 100% autonomous recovery even during a nationwide power outage. Utreras reported zero data loss, zero manual rebuilds, and complete confidence in pipeline performance. This freed up his team to focus on higher-value work like building predictive models and expanding to new markets.

With newfound confidence in Fivetran and its substantial connector library, he soon had each mining location centralize data into Databricks on Azure. It only took 5 months for Utreras’ team to deploy and replace 8 business-critical pipelines per site across all locations. Without Fivetran, this would likely have taken at least a year.

Expanding use cases to business operations data for faster decisions

With core operations stabilized, SQM began the next phase of its Fivetran deployment: scaling integration across business systems. The team set up pipelines for Salesforce CRM reporting and ServiceNow to track internal incidents at each site. For new global markets, they are adding new data sources and operational systems across regions and business lines before new mining operations begin. This gives leaders real-time, proactive visibility into performance, operations, and growth.

Fivetran’s deployment speed underpins SQM’s ability to expand. New pipelines that once took weeks or months now go live in under a day, often in just 15 minutes. The internal team manages every integration without expanding the team or relying on outside support. That agility has removed bottlenecks and accelerated experimentation across the business, helping SQM respond faster to changing conditions and new opportunities. According to Utreras, Fivetran has already had a substantial impact on SQM’s ability to do business.

Stakeholders across logistics, maintenance, chemistry, and supply chain now have dependable, up-to-date data to guide daily decisions. Yield tracking, equipment diagnostics, and transport coordination have all improved with better access to fresher insights. Data scientists are using timely data to build predictive models that optimize plant output and asset performance. Meanwhile, the Analytics & Digital team is building large language model (LLM)-based tools that allow internal users to query real-time data through natural language, which is an early step in SQM’s move beyond static dashboards.

“Fivetran provides the data reliability we need to confidently build LLM-powered tools, making real-time insights accessible across the business — no SQL required.”
— Jorge Utreras, Head of Projects, Architecture and Data Office at SQM

Supporting global growth with trustworthy data

As SQM expands into new territories, Fivetran continues to play a central role in supporting rapid growth. The company has launched new commercial operations in India and Brazil, each with site-specific systems that are connected and made operational from day one. The infrastructure behind that data can easily scale as fast as the business does.

Fivetran is enabling a unified data architecture across all locations, ensuring consistency and compliance by centralizing data in Databricks and making it accessible through Unity Catalog for governance and oversight. Regardless of the system’s origin — whether it’s a new commercial tool in Brazil, a CRM instance in India, or a maintenance database at a mining site — Fivetran standardizes ingestion, applies governance policies through Databricks Unity Catalog, and delivers data at the same near real-time cadence. The result is faster decision making, lower operational overhead, a leaner enterprise-wide data strategy, and — most importantly for SQM — reliability.

Looking ahead, Utreras and his team plan to expand their use of Fivetran to cover more applications like Salesforce and ServiceNow, accelerate the rollout of artificial intelligence (AI) and machine learning (ML) use cases, and deepen governance through Unity Catalog in Databricks. These efforts reflect a broader shift: from manual reporting and fragmented systems to an intelligent, automated data ecosystem — one that empowers faster decision making, accelerates global expansion, and supports cutting-edge AI initiatives across the business.

“Fivetran is essential to our global expansion strategy. It gives us fast, secure access to real-time data in new markets like India and Brazil, and makes it easy to connect new systems and deliver data from day one.”
— Jorge Utreras, Head of Projects, Architecture and Data Office at SQM

[CTA_MODULE]