IDC Research predicts that, by 2029, 60% of Global 2000 (G2000) enterprises will run most of their business processes AI-first.
The prediction— part of IDC FutureScape 2027, the company’s annual forecast for enterprise technology leaders—is predicated upon operations and decisions being left to AI agents.
IDC also predicts that by 2028, 70% of Global 2000 (G2000) enterprises will be able to continuously verify decisions their AI agents are making, up from 14% today.
IDC FutureScape 2027, titled The Great Rewrite: Winning in the Agentic Economy, comprises 250 predictions across 25 research areas, including AI factories, enterprise software, quantum computing, financial services, healthcare and government.
A dedicated team of IDC analysts in each area draws on the company’s proprietary research to identify the 10 developments that will shape that area, with projected timing and business impact. Each of the 25 sets of predictions was produced independently. Compared side by side, they reached the same conclusion: AI is rewriting every layer of enterprise IT at once, but the forces involved are altering priorities at different speeds. IDC sorted those forces by the situational window each one presents: some windows are closing; others are opening and yet a third category is just beginning to emerge.
“Nearly every tech leader I talk to wants to know whether they are behind on AI,” says Meredith Whalen, chief research and data intelligence officer at IDC. “The answer depends on the window. Some windows are closing, and more spending will not reopen them. Others are still open, but the time to shape them is shrinking.
“In every case, the winners will be the organisations that verify what each agent does as fast as they deploy it. IDC FutureScape 2027 puts a date and a number on each window so leaders can see which decisions can still change the outcome for their organisation.”
Inside IDC FutureScape 2027
The Closing Window – the advantage for companies with a head start is compounding:
- Rise of open models attacks the AI Economics challenge. By 2028, enterprises will shift 60% of AI workloads from proprietary LLMs toward open-source models and leverage model routing strategies to reduce costs and hedge against vendor lock-in. Enterprises that make those moves quicker will be the first to counter the agent inference overspend.
- Data volume outruns data value. IDC predicts 60% of enterprises will fail to convert their data into measurable business value by 2028, as data volumes outpace their ability to define and govern the value of that data.
The Open Window – the deadline to influence these fights is compressing:
- Governance cannot keep pace with agent velocity. Ten research areas, working independently, flagged the same gap: oversight is losing the race against agent deployment speed. IDC predicts G2000 enterprises without kill switches (controls that can shut down a misbehaving agent) for AI agents in production by 2028 will see 25% higher liability insurance premiums and twice as many material audit findings.
- Agent deployment is accelerating. In IDC’s Future Enterprise Resiliency and Spending Survey, 95,4% enterprises have deployed agents into an average 11 production workflows through July 2026, and their monthly inference budget is close to $118 000. The share of enterprises worldwide whose agentic bills have risen with no published price change will also grow, from 20% today to 55% by the end of 2027, according to IDC.
- Software contracts are being rewritten around agents. By 2028, 75% of G2000 vendor contracts will include liability clauses for agent-initiated actions, IDC predicts, making their absence a red flag in vendor selection.
- Global technology architecture is breaking apart. Nine research areas report that diverging regulation has made maintaining a single global architecture increasingly impractical. By 2029, 35% of G2000 companies will run jurisdiction-specific AI model registries and controls, IDC predicts.
The Emerging Window – new categories open to whoever moves first:
- A new job category is being created to supervise machines. By 2029, IDC predicts G2000 enterprises will shift 30% of entry- and mid-level execution roles toward agent supervision and orchestration. In IDC’s Future of Work Survey, 49% of midmarket organisations (1 000 to 4 999 employees) expect entry-level headcount to grow over the next 18 months, and 17% expect a decline.
- AI is leaving the screen. Eight research areas, including energy, manufacturing, retail and connectivity, describe AI moving into physical operations. By 2030, IDC predicts 40% of electric utilities will use graduated authority frameworks that let AI agents carry out selected routine grid-control actions, reducing operator intervention by 25%.
- Midmarket organisations are ahead of enterprises on agent deployment. In IDC’s Future of Work Survey of 1 146 business leaders, 46% of midmarket organizations (1 000 to 4 999 employees) have deployed agentic AI across several departments, ahead of 40% of enterprises (5,000 or more employees). Only 31% of those midmarket organizations have a fully operational AI center of excellence to oversee AI governance, compared with 43% of enterprises.