The AI Budget Gap: Why 20% of Tech Spend May Not Yield Firmwide ROI
AI-generated image · US National Wire
Forrester warns that a mismatch between technology investment and organizational change is leaving CIOs struggling to hit productivity targets.
For enterprise leaders, the current AI trajectory is creating a precarious financial tension. As The Register first reported, Forrester predicts that by next year, AI will consume 20 percent of all business technology spending. Yet, this aggressive capital allocation is colliding with a sobering reality: the technology is not yet paying for itself on a systemic level.
From an operational lens, the primary friction point is not the software, but the structure. A Forrester State Of AI Survey 2026 reveals that only 22 percent of AI decision-makers have seen a positive impact on their EBITDA over the last 12 months. The report argues that while companies are investing heavily in adoption, they are failing to implement the necessary changes to organizational structures, processes, and workflows required to drive measurable results across the firm.
This gap is creating a high-stakes environment for CIOs. According to The Register, Forrester expects over 50 percent of these executives to fall short of their AI-related ROI targets in the coming year. The pressure is intensified by the fact that 45 percent of AI decision-makers require a positive return on investment within a single year to deem their projects successful.
**Opinion: The Efficiency Pivot**
In my view, we are entering a phase of 'pragmatic optimization.' The era of blind investment in the most expensive frontier models is ending, replaced by a calculated effort to lower the 'AI tax' on the budget.
We see this shift in how firms are allocating their token usage. Forrester notes that companies are increasingly turning to open-weight models—such as Llama, Qwen, or Gemma—for high-volume, routine workloads. These include internal knowledge searches, customer support, document processing, and agentic workflow substeps. Forrester predicts these open-weight models will account for 50 percent of AI tokens while representing only 5 percent of total enterprise spend.
High-cost frontier models—specifically 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—are being relegated to a narrower set of 'critical task automations' where errors carry a high price tag.
As Sharyn Leaver, Forrester's chief research officer, told The Register, the challenge for leaders is now a balancing act: proving value while managing risk, infrastructure, and costs amidst geopolitical pressures and increasing operational complexity. Without a fundamental shift in how companies operate, AI will remain a series of isolated wins rather than a catalyst for firmwide productivity.