The AI Budget Bubble: Why Operational Change Must Outpace Spending
AI-generated image · US National Wire
Forrester warns that a fifth of tech budgets will flow into AI by 2027, but most CIOs are failing to link that spend to actual EBITDA growth.
For too long, the narrative around AI has been one of theoretical productivity. But as we move toward 2027, the industry is hitting a wall where the math simply isn't adding up. We are seeing a massive disconnect between the capital being deployed and the operational reality of the firms spending it.
As first reported by The Register, the analyst firm Forrester predicts that AI will consume 20 percent of all business technology spending next year. Yet, this surge in funding is not translating into broad financial wins. Forrester's State Of AI Survey 2026 reveals that only 22 percent of AI decision-makers have seen a positive impact on their EBITDA (earnings before income tax, depreciation and amortization) over the last year.
**Opinion:** The industry needs to stop treating AI as a magic bullet for productivity and start proving where the actual operational efficiencies are before the budget bubble bursts. Throwing money at frontier models without restructuring how a company actually functions is a recipe for failure.
Forrester notes that the primary hurdle is a lack of organizational evolution. The report argues that measurable, firmwide results require changes to processes, organizational structures, and ways of working. Without these shifts, Forrester expects success to remain confined to isolated use cases rather than comprehensive programs. This gap is creating a crisis of confidence: more than half of CIOs are likely to miss their AI-related ROI targets next year, and 45 percent of decision-makers require a positive return within a single year to deem their investments a success.
To mitigate these costs, enterprises are shifting their technical strategies. The Register reports that firms are increasingly turning to open-weight GenAI models—such as Llama, Gemma, or Qwen—for high-volume, routine workloads like internal knowledge search, customer support, and document processing. Forrester predicts these open-weight models will account for half of all AI tokens while representing only five percent of total enterprise spend.
High-cost frontier models will likely be reserved for critical tasks where errors are costly. These include Google's Gemini 4 Argon, OpenAI's GPT-6 Astra and GPT-6.1 Sol, and Anthropic's Claude Sonnet 5.5 and Claude Opus 5.5.
As Sharyn Leaver, Forrester's chief research officer, points out, the pressure is mounting for leaders to prove value while managing risk, infrastructure, and costs. In an environment of geopolitical pressure and increasing technical complexity, the 'spend first, figure it out later' approach is no longer sustainable.