Cloud Adoption in Brazil: AWS, Azure, GCP, and the Sovereign Question in 2026

Brazil’s public cloud market crossed $7.5 billion in 2025 and is on track for double-digit growth into 2026. The hyperscalers all operate São Paulo regions, with several adding Rio or other secondary regions. The sovereign-cloud question — increasingly contested elsewhere in the world — is mostly absent in Brazil, where the practical posture has been hyperscaler-first with specific carve-outs for federal-government workloads.

For enterprise buyers procuring cloud in Brazil, the practical landscape in 2026 is more straightforward than in some other markets but with its own specifics.

The hyperscaler shares#

AWS is the share leader with roughly 40-48% depending on methodology. Strongest in digital-native businesses, fintech (Nubank, Stone, Inter all run substantially on AWS), e-commerce, and increasingly mid-tier enterprise. The São Paulo region (sa-east-1) has been operational since 2011 and is the largest AWS region in Latin America. A second Brazilian region in Rio (sa-east-2) is in build-out.

Microsoft Azure is the second with 28-34%. Strongest in large enterprise with established Microsoft relationships, in public sector where Microsoft’s government go-to-market is deeper, and increasingly in AI workloads through Azure OpenAI. The Brazil South region (São Paulo) is the primary; Brazil Southeast (Rio) is the secondary.

Google Cloud is third with 10-15%. Strongest in data and AI-driven shops, in media and advertising, and in retail. The São Paulo region has been operational since 2017.

Oracle Cloud has a meaningful share in BFSI specifically — Brazilian banks running Oracle Database move workloads to OCI. Two regions (São Paulo, Vinhedo).

IBM Cloud and the domestic clouds — Embratel (Claro’s enterprise unit), Locaweb, Algar Tech — round out the remainder, with strongest position in mid-market enterprise and in specific regulated workloads.

Data residency and the sovereign question#

Brazilian regulatory data-residency requirements are lighter than India’s or China’s. The general framework:

LGPD does not require localization. Cross-border transfer is permitted under the specific mechanisms (covered in the LGPD post), but data does not have to remain in Brazil by default.

BCB has specific requirements for financial data — primary copies must be in Brazil for certain regulated entities, but the hyperscaler Brazilian regions satisfy this.

ANPD has not (as of mid-2026) issued country-restriction notifications that would prevent transfer to specific destinations.

Government data procurement uses specific cloud capacity, often through the SLTI and the various SaaS-via-government frameworks.

The result is that Brazilian enterprises generally use hyperscaler regions in Brazil for residency-sensitive workloads and the global hyperscaler footprint for everything else. The sovereign-cloud narrative that is dominant in some other markets is comparatively muted in Brazil.

Cost shape#

A few patterns specific to Brazil.

BRL pricing exposure. Hyperscaler pricing is generally USD-denominated; BRL fluctuations against USD have produced significant cost variance for Brazilian enterprises over 2022-2026. Many enterprise customers have negotiated BRL-denominated contracts or hedged exposure.

Reserved instance and savings plan discounts are similar to other markets — typically 25-50% off list for committed capacity.

Egress costs matter — cross-region traffic to and from Brazilian regions can be costly. Architects should plan for region-locality of services that talk to each other heavily.

Marketplace and partner channel. Brazilian enterprises often procure through partners — TIVIT, BRA cloud, Logicalis, Atos — for added operational support. The partner margin is real but the operational value is often worth it for complex environments.

Regulatory specifics#

A few regulator-specific cloud considerations:

BCB CSU (Cloud Service Usage) guidance for regulated financial entities requires specific risk assessment, exit-strategy documentation, and data-location commitments.

ANATEL has specific requirements for telecommunications data handling.

SUSEP for insurance data.

TCU (the federal accounts tribunal) has issued specific guidance for federal-government cloud procurement, which has shaped how the larger public-sector cloud deployments are structured.

The Estratégia de Governo Digital (the federal digital-government strategy) provides the umbrella framework for federal cloud adoption, and specific procurement vehicles route through it.

What’s working — and where the gaps are#

Working well: Mid-market enterprise cloud migration. Brazilian businesses have moved increasing portions of their portfolio to cloud, with strong AWS and Azure adoption.

Working well: Cloud-native startup and fintech infrastructure. Brazilian unicorns (Nubank, Inter, Stone, Loft, etc.) run substantially on cloud, mostly AWS.

Working but uneven: Large traditional enterprise. The Itaú-Bradesco-Petrobras tier has moved more slowly, partly because of legacy infrastructure complexity, partly because of integration with existing data center investments.

Working but uneven: Government. Federal cloud adoption has progressed under the digital-government strategy; state and municipal cloud adoption varies.

Not yet working: Cloud-resident sovereign AI. The Brazilian sovereign AI infrastructure question is still emerging; most AI training continues to use US or EU cloud regions.

What’s coming in 2026 and 2027#

Three things to watch:

Hyperscaler AI infrastructure in Brazilian regions has been expanding — GPU availability has historically been a constraint, with most production training running in US regions. The 2024-2026 hyperscaler buildout in Brazil has improved this.

The MERCOSUL data-flow harmonization efforts may produce simpler cross-border data flow within the bloc.

The federal cloud procurement framework is being refreshed with implications for how the major federal contracts are structured.

Procurement patterns we recommend#

For a Brazilian enterprise doing cloud procurement in 2026:

1. Pick a primary hyperscaler based on workload fit and partner ecosystem. AWS for digital-native, Azure for Microsoft-anchored, GCP for data/AI-led shops.

2. Negotiate BRL pricing where possible. Cross-currency exposure has been material.

3. Plan for the Rio secondary region for the hyperscalers that have it. Multi-region resilience in Brazil is now operationally workable.

4. Engage partner channel deliberately. Brazilian enterprise cloud procurement through partners is common and often valuable.

5. Monitor egress aggressively. Brazilian-to-US egress costs add up.

6. Build LGPD compliance into the cloud architecture from the start.

Where pdpspectra fits#

We run cloud architecture and migration programs for clients in Brazil and broader LATAM. Our work is hyperscaler-pragmatic and accounts for the specific Brazilian regulatory and procurement context.

Related reading: the India cloud adoption post, the Brazil LGPD post, and the Snowflake vs Databricks vs BigQuery post.

---

Brazil’s cloud adoption is mature and accelerating. Talk to our team about your architecture.