Average reasoning token consumption per enterprise organization climbed approximately 320 times over the past 12 months, unveiling a massive, often hidden, surge in AI operational costs, according to PYMNTS. The exponential increase in reasoning token consumption confirms widespread AI tool adoption, directly escalating costs for data processing and response generation.
Companies are dramatically increasing AI budgets and consumption, yet a significant portion fails to meet promised cost-saving targets. The disconnect between increasing AI budgets and failing to meet cost-saving targets generates tension between investment ambition and measurable financial returns, especially as AI integration's impact on enterprise IT spending in 2026 clarifies.
Enterprises risk spiraling IT costs, increased vulnerability to AI-enabled attacks, and diminished shareholder value. The risk of spiraling IT costs, increased vulnerability to AI-enabled attacks, and diminished shareholder value persists if enterprises fund AI initiatives without rigorous ROI validation and enhanced cybersecurity measures.
Key AI Spending Statistics for 2026
- 320 times — Average reasoning token consumption per enterprise organization increased over the past 12 months, according to PYMNTS.
- 90% — Of companies are increasing their AI budgets again, despite not meeting prior cost-saving targets, according to Bain & Company.
- 44% — Of companies are funding the next wave of AI investments from savings from prior automation programs that have consistently come in below target, according to Bain & Company.
- 7% — Of companies are running fully autonomous AI agents in production today; the dominant model still requires human approval, according to Bain & Company.
- 72% — Increase in AI-enabled attack activity is projected for 2025, according to Splunk.
The AI Spending Paradox: More Investment, Less Return
Nearly 40% of companies measuring AI cost savings landed below 10%, despite targeting 11% to 20%, according to Bain & Company. The shortfall in AI cost savings confirms a consistent failure to achieve anticipated financial efficiencies from AI deployments. The fact that 90% of companies are increasing AI budgets despite not meeting prior cost-saving targets reveals a market driven more by FOMO and vendor pressure than by demonstrable ROI.
Further, 44% of companies fund new AI investments from savings from prior automation programs that consistently underperformed, according to Bain & Company. Funding new AI investments from savings from prior underperforming automation programs establishes a self-defeating investment cycle. Underperforming initiatives cannibalize funds for new, potentially equally underperforming, AI projects. Enterprises are caught in escalating AI spending and consumption, driven by perceived necessity, yet consistently fail to realize promised cost efficiencies.
| Metric | Observation | Implication for Enterprise IT Spending |
|---|---|---|
| AI Cost Savings Achieved | Nearly 40% of companies landed below 10% (Target: 11%-20%) | Significant gap between expected and realized cost efficiencies, leading to budget overruns. |
| Companies Increasing AI Budgets | 90% are increasing budgets, despite prior shortfalls | Investment decisions are driven by factors beyond proven ROI, potentially escalating unsustainable spending. |
| Funding New AI Investments | 44% use savings from prior underperforming automation programs | A cycle of reinvesting into unproven initiatives, hindering overall financial performance. |
| Average Reasoning Token Consumption (12 months) | Climbed approximately 320 times per organization | Massive increase in operational costs for AI, often hidden from initial budget projections. |
Data compiled from Bain & Company and PYMNTS.com.
Driving Forces: Market Pressure and Limited Autonomy
Apptio is launching a preview of its new Conversational Insights solution, an enterprise-grade conversational interface with AI embedded across its solutions portfolio, according to IBM Newsroom. Apptio's launch of its Conversational Insights solution demonstrates pervasive market pressure for enterprises to adopt AI, as vendors embed AI into existing software solutions. This widespread availability pushes companies to allocate resources towards AI capabilities, irrespective of immediate, quantifiable returns.
However, only 7% of companies run fully autonomous AI agents in production today; the dominant model still requires human approval, according to Bain & Company. The 320-fold increase in average reasoning token consumption per enterprise organization (PYMNTS.com), coupled with only 7% autonomous AI agents (Bain & Company), confirms enterprises incur massive operational costs for AI still largely requiring human oversight. The massive operational costs for AI still largely requiring human oversight undermines automation's promise. The pervasive integration of AI into existing enterprise solutions and the sheer scale of overall IT spending create immense pressure for adoption, even as the technology's full autonomous potential and efficiency gains remain largely unrealized.
Industry projections for overall IT spending also present a complex picture for AI's contextualization. Splunk projects global IT spending to exceed $6 trillion in 2026, while hginsights projects it to reach $4.96 trillion in the same year. The significant discrepancy (over $1 trillion) between Splunk's and hginsights' projections for global IT spending reveals a lack of consensus on the overall market size, making it difficult to accurately contextualize AI's share or impact on total IT budgets.
The Shifting Landscape: New Winners, Growing Risks
Global cybersecurity and risk management spending is expected to grow 12.5% to $240 billion in 2026, according to Splunk. The expected 12.5% growth in global cybersecurity and risk management spending directly aligns with the rapid adoption of AI. The benefits of AI adoption are immediately offset by a new wave of cybersecurity threats.
AI-enabled attack activity increased by 72% in 2025, according to Splunk. The 72% increase in AI-enabled attack activity forces companies into a reactive and costly defense posture. The 72% increase in AI-enabled attack activity (Splunk) confirms that AI adoption benefits are immediately offset by new cybersecurity threats, compelling companies into a reactive and costly defense. While AI solution providers and cybersecurity firms see significant growth, enterprises face escalating security threats and the burden of justifying massive, often underperforming, AI investments against ever-increasing IT budgets.
If enterprises do not pivot from reactive AI investment to strategic implementation with rigorous ROI validation and enhanced cybersecurity, their IT budgets will likely inflate further without commensurate value, exacerbating financial strain and security vulnerabilities.
