Petrobras HPC Strategy 2025: How In-House Supercomputing Creates a Competitive Moat

Petrobras’s HPC Projects: From R&D Tool to Strategic Pillar

Petrobras shifted its digital strategy from acquiring computational capacity to building a proprietary, vertically integrated High-Performance Computing (HPC) and data infrastructure to secure its data and accelerate model development.

• Between 2021 and 2024, Petrobras focused on acquiring supercomputers like Pégaso and its first AI-specific machine, Tatu, establishing a powerful computational base to support its digital transformation goals, which included a $2.1 billion innovation budget.
• Starting in 2025, the strategy evolved into creating a dedicated ecosystem with the landmark R$2.3 billion (US$500 million) contract with Elea Data Centers for a private AI data center, signaling a move towards long-term data sovereignty and in-house control.
• This shift is validated by further acquisitions in 2025, including an Atos Bull Sequana supercomputer with advanced NVIDIA GPUs and the activation of the Lenovo ‘Harpia’ HPC, demonstrating a commitment to owning the entire AI stack from hardware to software.

Investment Analysis: Petrobras’s Multi-Billion Dollar HPC Commitment

Petrobras is committing hundreds of millions of dollars to proprietary HPC and data center infrastructure, a capital-intensive strategy designed to create a long-term competitive advantage through data control and model development.

• The cornerstone of this strategy is the R$2.3 billion (US$500 million), 17-year contract with Elea Data Centers announced in October 2025, which establishes a dedicated AI data center to house its supercomputers and critical workloads.
• In 2023, the company laid the groundwork by activating its first AI-specific supercomputer, Tatu, at a cost of BRL 36 million ($6.8 million), as part of a larger $2.1 billion digital transformation budget for the 2023-2027 period.
• The investment momentum continued into 2025 with the activation of the R$435 million (USD$25 million) Lenovo “Harpia” HPC and a separate 150 million reais (US$28 million) contract with Lenovo for general IT hardware, solidifying its infrastructure base.

Table: Petrobras’s Key HPC and Data Infrastructure Investments

Partnership Analysis: Petrobras’s Strategic Alliances for HPC Infrastructure

Petrobras is building its HPC ecosystem by forging long-term, strategic alliances with specialized hardware manufacturers and data center operators, reducing reliance on general-purpose cloud providers.

• The most significant partnership is the 17-year agreement with Elea Data Centers initiated in October 2025, which secures a physical home for Petrobras’s critical HPC and AI workloads and ensures data sovereignty.
• In May 2025, Petrobras partnered with Atos Bull and NVIDIA to acquire a new AI supercomputer, ensuring direct access to cutting-edge processing technology required for large-scale model development.
• The company established a recurring partnership with Lenovo, demonstrated by the 2025 activation of the “Harpia” supercomputer and a separate $28 million hardware contract, making Lenovo a key supplier for its digital infrastructure.
• Before 2025, partnerships with firms like Intel to create a “Deep Lab” and with IBM for the Cérebro platform established the foundational expertise for leveraging HPC in specialized applications like seismic analysis.

Table: Petrobras’s Key HPC and Technology Partnerships

Geography: Petrobras’s Brazil-Centric HPC Strategy

Petrobras’s HPC and data infrastructure strategy is hyper-focused on Brazil, designed to process and secure proprietary geological data from its domestic pre-salt assets.

• From 2021 to the present, all major HPC infrastructure investments, including supercomputers and data centers, are located within Brazil, reinforcing the strategy of keeping its most valuable data assets sovereign.
• The new Elea AI Data Center, a $500 million project announced in 2025, will be built in São Paulo, creating a central hub for the company’s national data processing needs and solidifying its physical infrastructure footprint.
• Research partnerships that leverage this HPC power, such as the one with the Federal University of Ceará (UFC), are also concentrated in Brazil to solve geological challenges specific to the Brazil Equatorial Margin.

Technology Maturity: Petrobras’s Path to Commercial-Scale HPC

Petrobras has rapidly advanced its HPC capabilities from an R&D and operational support tool to a commercially scaled, enterprise-wide strategic asset.

• Between 2021 and 2024, Petrobras’s HPC technology was in an application and expansion phase, with the activation of supercomputers like Pégaso and Tatu to support specific E&P tasks and accelerate R&D projects for seismic modeling.
• The period from 2025 to the present marks a shift to commercial-scale integration, evidenced by the $500 million investment in the dedicated Elea AI Data Center designed to support all enterprise workloads, from seismic analysis to generative AI.
• This maturity is validated by the strategic goal of leveraging this infrastructure to dominate reservoir characterization, moving beyond incremental efficiency gains to redefine operational standards in deepwater exploration.

SWOT Analysis: Evaluating Petrobras’s HPC Strategy in 2025

Petrobras’s primary strength is its integrated strategy of combining vast, proprietary pre-salt data with massive in-house HPC infrastructure, creating a powerful competitive advantage that is difficult to replicate.

• Strengths are rooted in the ownership of both unique geological data and the computational power to analyze it, a combination difficult for competitors to replicate.
• Weaknesses involve the high capital expenditure and long-term commitment required for building and maintaining proprietary data centers, creating potential financial inflexibility compared to rivals using outsourced cloud models.
• Opportunities lie in leveraging this integrated stack to achieve superior operational efficiency, lower production costs, and develop new AI-driven services that can be commercialized.
• Threats include the rapid pace of technological change in HPC and AI, which could make current investments obsolete, and competition from more agile, cloud-native strategies pursued by rivals.

Table: SWOT Analysis for Petrobras’s HPC Strategy

Forward-Looking Outlook: From Hardware to Enterprise AI

The next strategic step for Petrobras is to fully operationalize its massive HPC infrastructure to scale its AI models across the entire enterprise, moving from project-based ROI to systemic operational transformation.

• The 2025 investments in the Elea Data Center and the Atos Bull supercomputer indicate the hardware foundation is now largely in place, shifting the strategic focus toward software development and widespread deployment.
• Success will be measured by how effectively Petrobras uses this in-house power to scale tools like Chat Petrobras and the NVIDIA Ne Mo knowledge assistant to its 110, 000 employees and expand AI analytics beyond E&P.
• The ability to translate the raw power of its new supercomputers into quantifiable outcomes, such as further improvements on the 50% accuracy gain in geological models, will determine the ultimate success of this multibillion-dollar strategy.

Frequently Asked Questions

Why is Petrobras investing in its own supercomputers and data centers instead of using public cloud services?

Petrobras is building its own proprietary HPC and data infrastructure primarily to secure its vast, unique geological data and achieve long-term data sovereignty. This in-house control allows the company to accelerate the development of proprietary AI models for exploration and production, creating a competitive advantage that is difficult for rivals to replicate.

What is the single largest investment in Petrobras’s new HPC strategy?

The cornerstone of the strategy is the R$2.3 billion (US$500 million) contract with Elea Data Centers. This 17-year agreement, announced in October 2025, is for building a dedicated AI data center in São Paulo to house Petrobras’s supercomputers and critical workloads.

How does Petrobras’s in-house HPC infrastructure create a ‘competitive moat’?

The competitive moat comes from the unique combination of owning two critical assets: vast, proprietary pre-salt geological data and the massive, dedicated computational power to analyze it. This integrated stack allows Petrobras to achieve superior operational efficiency, lower production costs, and develop advanced AI-driven solutions for deepwater exploration faster than competitors who may not own their entire infrastructure.

Who are the key technology partners helping Petrobras build its HPC ecosystem?

Petrobras is forging strategic alliances with specialized vendors. The most significant is the 17-year partnership with Elea Data Centers. For hardware, it works with Atos and NVIDIA for its latest AI supercomputers (featuring Hopper and Blackwell GPUs) and has a recurring partnership with Lenovo for supercomputers like ‘Harpia’ and general IT equipment.

What are the main risks associated with Petrobras’s HPC strategy?

The primary risks involve the high capital expenditure and financial inflexibility from its long-term, multi-billion dollar commitments, such as the 17-year data center contract. Another major threat is the rapid pace of technological change, where its massive investment in current-generation hardware could become obsolete, especially compared to rivals pursuing more agile, outsourced cloud strategies.