Chapter One

Cloud was the Strategy: AI Changed the Question

The New Test of Cloud Strategy

A cloud strategy that cannot explain AI workload placement, data control, agent governance, economics and talent implications is no longer a complete strategy.

For years, cloud sat at the center of the digital transformation agenda. It provided a unifying narrative: exit fragmented estates, consume modern platforms, scale on demand, and accelerate product delivery.

The operating assumption was clear: cloud first, then optimize.

AI disrupts that sequence. It brings new demand patterns – large-scale training, unpredictable inference, vector search, high-volume data movement, model observability, agent orchestration and continuous evaluation.

It also brings new governance concerns: who can send data to which model, who approves autonomous actions, who pays for token-intensive workflows, and who is accountable when an AI-enabled decision fails.

The board question is therefore changing from "How fast can we get to cloud?" to "Have our technology choices made us more or less ready for AI?" That is a sharper test.

It asks whether infrastructure is elastic and economical, data is accessible but protected, operating models are product-centric, and executives can see where value is created and where cost is leaking.

_{Traditional front and backend environmental carbon footprint monitoring and measurements are already mature, and AI rating models are close behind. Industry standard energy-label-style rating (A-F) for AI models based on sustainability is being developed and will allow CIOs to ensure ESG targets are met while balancing productivity benefits.}

Steve Crombie, CIO Bulgaria, ING Bank

AI Exposes Seven Unresolved Tensions

Speed vs Control

Business teams want rapid experimentation; risk functions need traceability, security and enforceable policies

Scale vs Economics

AI can unlock productivity, but inference, data movement and embedded SaaS AI features can create usage-based cost volatility

Ecosystem Leverage vs Sovereignty

Hyper-scalers and SaaS vendors can accelerate adoption, but overdependence may weaken bargaining power and strategic flexibility

Automation vs Accountability

Autonomous systems can take work out of the process, but they cannot take accountability out of the enterprise

Capability Velocity vs Architectural Stability

Models, frameworks and the surrounding tooling change faster than enterprises can standardize, so any architecture decision risks being stranded by the next generation of capability before it has paid back

Ambition vs Data Readiness

AI strategy is constrained less by access to models than by the state of enterprise data - quality, lineage, classification and access - and most programs stall on data foundations long before they stall on model capability

Direct Resource Consumption vs Benefit to Societal Use Cases

AI is environmentally expensive and requires evaluation of benefits to society, against traditional cloud solutions with less environmental impact

_{Cloud solved where workloads run. AI forces us to rethink how decisions are made, how knowledge is accessed, and who remains accountable when work is increasingly performed by digital agents.}

Sumita Dutta, Global IT and Digital Head & CIO, Shell Downstream, Customer Operations, Shell