Who is Fernando F. Azevedo?

Fernando F. Azevedo is a Senior Solutions Architect at Banco Itaú with 16+ years of experience across AWS, event-driven architecture, DevSecOps, Data Mesh, AI and financial systems.

What technical topics does Fernando work with?

Fernando works with AWS, Kubernetes, Kafka, Data Mesh, Amazon Bedrock, RAG, DevSecOps, observability, financial systems and architecture communication using C4, ADRs and trade-off analysis.

Is Fernando available for professional conversations?

Fernando is currently building at Banco Itaú and is open to thoughtful conversations about architecture, cloud, AI, engineering leadership, community, podcasts and technical collaboration.

Technology

Transition from Physical Infrastructure to AI: A New Landscape

Jun 16, 2026 5 min AI-assisted

ARCHITECTURE

Architecture visual generated from the article theme.

Reading

Estimated time

Pillars

Well-Architected

Topics

Connected themes

Good morning! Today, we'll discuss how physical infrastructure is evolving to make way for artificial intelligence. We'll analyze Google's investment in Alabama and other key developments.

What you will take away

How to connect IA and infraestrutura to real architecture decisions.

Which trade-offs to watch before scaling the solution.

How to reason about security, cost, operations and reliability together.

A practical checklist to turn insight into execution.

The Shift in Focus of the Technology Industry

Good morning, readers. Today, in Morning Technology, I bring an updated panorama on the industry's shift from physical infrastructure to artificial intelligence (AI).

Google's Announcement

The starting point is Google's recent announcement. The company announced a $1.5 billion investment to expand its data center campus in Jackson County, Alabama, for the years 2026 and 2027. This investment is not just about physical expansion but also about supporting the local community.

Physical Infrastructure vs. Scaled Computing

This news highlights the importance of physical infrastructure in the era of AI. Without energy, land, network, and physical capacity, it is impossible to sustain the scale of computing necessary for AI. The race for AI is no longer just about creating better models, but about who can provide the necessary infrastructure to support them.

Controls and Governance in AI

Another crucial point is the evolution of controls and governance in AI. The release of version 2.1.178 of Anthropic's Claude Code model, for example, brings significant improvements in this regard. More granular control and governance are essential to ensure that AI is used safely and ethically.

Distribution and Corporate Channel

AI distribution via browser and corporate channels is also gaining prominence. This approach not only facilitates access but also allows companies to integrate AI into their business processes more effectively.

The Silence of OpenAI

Finally, it's worth mentioning the strategic silence of OpenAI. While other companies announce their plans, OpenAI maintains a more discreet profile. This may indicate a different strategy, but it also raises questions about the future of its initiatives.

Conclusion

In summary, the technology industry is undergoing a significant transformation. Physical infrastructure is evolving to make way for artificial intelligence, and companies are investing heavily to ensure they can sustain this scale of computing. More rigorous controls and distribution via browser and corporate channels are essential parts of this new phase.

I hope you enjoyed this panorama. See you in the next edition!

Reference architecture

A reference view to organize ingestion, processing, governance and consumption for the discussed domain.

Recommended logical flow

Signals and requirements

Capture functional requirements, constraints, events, risks and business objectives.

Cloud-native platform

API Gateway / Lambda / EKS

Use a managed and observable layer to reduce operational load and accelerate evolution.

Data and event backbone

S3 / DynamoDB / MSK

Model data, events and retention with security, traceability and predictable cost.

Operate and improve

CloudWatch / Bedrock / CI/CD

Close the loop with metrics, automation, feedback and continuous improvement.

Implementation checklist

Practical items to turn the analysis into an execution plan.

Define objective and metric

Connect the theme to a business, operations or risk metric.

Map integration and data

List sources, consumers, sensitive data and trust boundaries.

Create guardrails

Implement limits, observability, security and cost control from the start.

Validate in controlled production

Use progressive rollout, alarms, failure testing and clear rollback.

Anti-patterns to avoid

Treating Technology as an isolated tool without process, metrics and operational ownership.
Scaling before defining limits, telemetry, unit costs and recovery strategy.

AWS Well-Architected lens

A pillar-based read that turns architecture decisions into sustainable operations.

Operational excellence

Define metrics, runbooks and automation from the initial design.

Security

Classify data, protect boundaries and avoid exposing secrets in automation or prompts.

Reliability

Consider service quotas, timeouts, retries and graceful degradation.

Performance efficiency

Choose managed services and validate bottlenecks with real tests.

Cost optimization

Tie cost to a unit of value and automate consumption alerts.

Sustainability

Avoid unnecessary processing and reduce purposeless retention.

#IA#infraestrutura#tecnologia#Google#Anthropic#OpenAI

References and next steps

Useful links to deepen the architecture decision.

AWS Architecture Center AWS Well-Architected Framework AWS Architecture Blog

Subscribe to the newsletter

Daily and weekly digests on AI, AWS, tech and markets. No spam, unsubscribe anytime.