AI in Business: The Platform Becomes Crucial by 2027

The shift towards inference, the weight of tokens, and risk management are pushing companies to reconfigure their infrastructure choices. Amid political constraints on data centers, hardware shortages, and scarce skills, platform teams are gaining a central role. Forecasts from Forrester and Gartner indicate increased internalization and budget pressures.

Local Opposition and Shortages Weigh on Infrastructure Projects

The 2026 midterm elections highlighted persistent public opposition to large data center projects. Initiatives aimed at bringing AI workloads on-site will have to contend with this resistance. Meanwhile, storage teams faced a rise in memory prices and a shortage of new capacities in 2026, tensions that are expected to continue influencing pricing for IT buyers in 2027. Even when funding is available, internal infrastructure plans also encounter skill shortages. The human investment required to operate an AI platform is substantial, and not all organizations can absorb these costs equally.

Inference Redistributes Loads and Favors Distributed Architectures

The transition from training to inference will fundamentally alter the design requirements for infrastructures in 2027. Training requires sustained throughput between computing, storage, and network that are closely synchronized, which has favored the rise of large AI data centers. In contrast, inference can run on more accessible hardware and requires bringing workloads closer to users to minimize latency, a particularly pressing need in industry and robotics. Specialists anticipate that inference will become the engine of self-managed AI infrastructures by 2027, supported by more diverse chips and deployed across more distributed sites.

Budgets Under Pressure: Tokens and Governance in the Spotlight

The proliferation of agents increases risks and drives up token-related bills. These costs already complicate the trade-offs between developing or buying, and are expected to rise with the widespread adoption of agents. Gartner estimates that by 2027, companies automating their IT operations with variable, uncapped pricing models tied to token usage will exceed their initial forecasts by more than 50%. Beyond spending, management must also consider the overall profitability of investments and the potential financial consequences of an agent acting without authorization.

The Internal Platform Becomes the Orchestrator of AI Workloads

In light of increasing usage, platform engineers will need to choose the execution environment for a growing number of workloads, assigning each to the most cost-effective model or agent service, whether it be a public cloud or on-premises infrastructure, while ensuring performance, resilience, security, and governance. After ten years of sporadic appearances from cloud-native and DevOps, many organizations now have fragmented platforms across teams and tools. According to Torsten Volk (Omdia), this fragmentation must fade as the risks of agentic AI increase, placing platform teams at the center. Leaders will seek more control and less expensive alternatives to cutting-edge and hosted offerings. In this context, the ability to centralize resource management across heterogeneous environments is seen as crucial for integration and operational efficiency.

Internalization and Executive Risks: Predictions for 2027

Forrester anticipates that 30% of large companies will build a dedicated private AI infrastructure for inference as early as next year, driven by rising risks and the protection of sensitive data. This internalization trend follows two years of acclimatization to language models and agents. On the executive front, Scott Sinclair (Omdia) predicts that by 2027, at least one CEO of a Fortune 500 company will leave their position or be ousted for failing to deliver a positive return on their AI investments.