AI Spending Rose 110% and the Underlying Systems Could Not Keep Up
ServiceNow has been measuring AI maturity in large enterprises for three years. In 2026, the index they construct through surveys of more than 4,500 executives and 2,000 employees worldwide recorded a notable improvement: 16 points of progress, reaching 51 out of 100. That sounds positive until you read the next line: while corporate AI spending grew 110% year-over-year, the foundational capabilities that AI needs to function at scale simply did not keep pace. The money arrived. The architecture did not.
That is the figure that should concern any CFO who approved an AI budget in the last 18 months without also reviewing the data, workflows, and governance protocols sitting beneath it. This is not a technology problem. It is a sequencing problem.
When Investment Precedes Structure
There is a pattern that repeats itself with every major technology wave: spending gets ahead of readiness. It happened with ERP, it happened with the cloud, it happened with the big data programs of the mid-2010s. The difference now is scale. According to Gartner projections, global AI spending will reach $2.59 trillion in 2026, a 47% increase over the previous year. Stanford HAI documented that global private investment in AI reached $344.7 billion in 2025, a 127.5% increase over 2024. And in the first quarter of 2026 alone, Amazon, Google, Microsoft, and Meta jointly invested $130.65 billion in infrastructure, largely directed toward AI.
Those numbers are real. The problem lies in what they do not measure: whether the companies fueling that demand have the internal systems to take advantage of what they are buying.
The ServiceNow index pinpoints precisely where the breakdown occurs. Organizations purchased AI models, tools, and platforms. But their data remains fragmented in silos that do not communicate with one another, their workflows were not redesigned before being automated, and their employees were not brought into the change process with sufficient lead time. The structural consequence is predictable: AI delivers inconsistent results, teams lose confidence in the system's outputs, and projects stall indefinitely in pilot phases without ever reaching production.
Holly Briedis, Senior Vice President of Global Industries and Solutions at ServiceNow, describes it with a precise image: most organizations are trying to run Formula 1 on go-kart infrastructure. Data operates as a disconnected patchwork, and when you attempt to run horizontal workflows across that environment, the seams become visible immediately.
Those Who Are Advancing Made a Decision First
The index identifies a group it calls Pacesetters: the 21% of surveyed organizations with the highest maturity scores. What distinguishes them is not that they spent more, but that they integrated and modernized their data before deploying AI, not after. Sixty-four percent of this group integrates and optimizes their data digitally, compared to only 14% of the rest. They do not encounter fewer data problems than everyone else; they resolve them faster because they built the capacity to detect and correct those problems as part of the system's design.
They also differ in how they understand the purpose of AI within the business. Fifty-seven percent of Pacesetters establish a shared strategic vision for AI that goes beyond operational efficiency. They are not automating broken processes — they are redesigning how work flows. The distinction is not semantic: automating a poorly designed process only produces errors faster and at greater scale.
There is an implicit decision embedded in this approach that deserves to be named explicitly. Pacesetters effectively sacrificed deployment speed in the short term. They committed to building the foundations before raising the floors. In the current environment, where boards of directors demand AI use cases in every quarterly presentation, that sacrifice carries a considerable internal political cost. Those who sustained it are now reaping an advantage that is not easily replicated in six months.
Governance Is Not a Brake — It Is What Enables Scale
One of the most significant findings in the index speaks directly to leadership teams: half of the surveyed employees believe their jobs will become less necessary as agentic AI evolves, and they do not feel their organizations are preparing them for what is coming. Briedis names it without euphemism: that is not a skills gap, it is a leadership gap.
The point has a concrete mechanism. When employees do not trust the outputs that AI delivers, they will work around the technology. They will create their own shortcuts, their own workarounds, their own parallel records. The AI deployment is formally sustained on paper, but operationally it is hollowed out. The return on investment never materializes — not because the technology failed, but because the organization never truly adopted it.
Well-designed governance operates in exactly the opposite way from how it is typically presented. It is not a set of restrictions that slows AI down; it is the scaffolding within which the system's autonomy can operate without generating mistrust. When protocols are clear, teams can distinguish between a reliable output and one that requires review. When they are not, the human default is to distrust everything, and the technology ends up underutilized.
The same applies to data. While a person can work around an information silo by relying on their judgment, contextual knowledge, and the informal connections accumulated over years of work, AI cannot. AI was designed to solve problems horizontally, cutting across functions and departments. If data does not flow between those functions, AI operates with a fraction of the information it needs and delivers a fraction of the value it could deliver.
What the Gap Between Spending and Maturity Reveals
The pattern documented by the ServiceNow index is not new, but it has now reached a scale where the consequences are more difficult to absorb. Oxford Economics projects that global AI spending will grow from $340 billion in 2025 to approximately $3 trillion by 2035, representing roughly 23% of total enterprise technology spending. BCG documents that corporations expect to double their AI spending in 2026, with AI rising from 0.8% to 1.7% of their revenues.
At that magnitude of investment, the cost of not having the underlying systems prepared ceases to be an IT problem and becomes a capital allocation problem. If spending doubles but operational maturity does not advance in the same proportion, companies are building a growing financial burden on a foundation that is still not capable of supporting it. Pilots that never reach production are not neutral learning failures: they consume budget, the time of technical teams, and above all, the internal credibility required for the next investment cycles.
The sectors showing the greatest urgency are those with the most to lose in the event of falling behind. Technology companies and financial institutions plan to allocate around 2% of their revenues to AI in 2026, according to BCG. In those sectors, competition does not wait for laggards to get their data in order. The window between an investment decision and a visible result is no longer wide enough to remedy in parallel what should have been built beforehand.
The ServiceNow index, read as a sectoral diagnostic document rather than as a marketing piece, describes a particular moment on the enterprise AI adoption curve. Spending has already committed the bet. What determines who converts that bet into a sustainable advantage is not the next model that gets purchased or the next pilot that gets announced: it is whether the data architecture, the redesigned processes, and the organizational culture can sustain what the budget has already promised to deliver.
