AI Transformation Is a Political Minefield, Not a Technology Project
(SeaPRwire) –
By: Christian Pierce
Few Fortune 500 companies have truly reinvented their core operations around artificial intelligence. That stark reality came straight from Marc Warner, Accenture’s global chief technology officer. He delivered it during a Brainstorm AI virtual event, and the number he gave was blunt: close to zero. The companies that have done it are outliers. The rest are layering AI onto workflows that were never designed to run differently.
What Warner called “politics projects” is really a structural problem dressed up as a technology one. Most enterprises have placed AI in peripheral work — marketing copy, meeting transcription, basic coding assistance. The returns there are measurable, incremental, and safe. But the actual economic transformation everyone promises lives in the core. That is where the resistance is. Greg Ulrich of Mastercard put it plainly. Speeding up one step in an 80-step process just creates a bottleneck downstream. An AI agent automating a single function is not the long pole in the tent. The real work is remapping the entire flow.
Matthieu Caillat at AXA admitted the company started with roughly 400 AI use cases to build internal enthusiasm. They had to narrow focus. The current trajectory projects between €500 million and €700 million in annual pre-tax recurring value from AI by 2029. That target is anchored against a backlog of about four billion unstructured documents the insurer is only now beginning to tap. The lesson is not about AI capability. It is about organizational capacity to absorb it.
The accountability gap is another structural failure point. When an AI agent makes a wrong decision, no one owns it cleanly. Warner laid out the three-way split: the business owner takes responsibility for value, the technologist for functionality, and the CISO for security. Nobody has solved the politics of large organizations. Mastercard runs weekly CEO update sessions with direct reports to force alignment. Even that does not guarantee outcomes.
Marjorie Janiewicz of Mistral AI echoed the frustration. After a decade in cybersecurity, she sees the real question as whether a company controls its own technology stack. Can it set its own guardrails? Open-weight models change the calculus slightly. But control requires infrastructure investment and skilled people. Most enterprises lack both.
Plans expire faster than ever. Ulrich noted that consulting tools Mastercard built a year ago are now quaint and useless. They are being rebuilt. Caillat said AXA used to measure twice and cut once. They cannot afford that luxury anymore. Conclusions expire before certainty arrives. The pace of change, according to Accenture’s calculations, shows no sign of slowing until around 2030. Energy constraints in the West may start to bite in the early 2030s, Warner warned.
The fundamental insight here is not about AI capability or computational power. It is about organizational metabolism. Companies that treat AI as a tool to bolt onto existing processes will get marginal gains. Companies that redesign their workflows from scratch around what AI enables will see transformation. The first group is vast. The second is nearly empty.
Warner urged most organizations not to dwell on blame right now. People need credit for trying things, knowing many will fail. There is an exception for critical incidents like major security breaches. But the broader point stands: the industry is still in the experimentation phase. The real restructuring has barely begun.
The end-game is clear. Enterprises that survive this period will be those that treat AI transformation as a political project first and a technology project second. The ones that do not will accumulate a graveyard of pilot programs and wasted spend. The tools are ready. The organizations are not.
Author bio: Christian Pierce is a chief financial columnist and markets commentator covering enterprise technology investment and organizational transformation trends.