AI's real test: governance beats speed, say Teneo & Thoughtworks CEOs

The article, authored by the CEOs of consulting firm Teneo and software services firm Thoughtworks, diagnoses a critical turning point in corporate AI adoption. Over the past year, large language models have become deeply embedded across business processes, shifting AI from an experimental tool to a fundamental part of operations. However, this embedding has exposed two structural problems that will now shape the competitive landscape.

The first problem stems from a faulty assumption about AI's economic impact. Many executives believed AI would allow them to swap human labor costs for model inference costs and capture the efficiency gain as profit. This thesis led to widespread layoffs. But the reality has proven more complex. Organizations that reduced headcount in anticipation of AI-driven productivity gains discovered they had eliminated institutional knowledge and engineering expertise critical to integrating and refining AI systems. Ford provides a concrete example: the company publicly stated it must rehire hundreds of engineers to fix quality control issues with newly implemented AI tools, directly citing the loss of veteran expertise as the cause. This pattern is hitting many companies that conducted large-scale layoffs.

The second problem is a misalignment between investor expectations and executive reality. Teneo's most recent annual CEO and investor survey revealed a stark gap: 53 percent of investors expected return on investment from AI within six months, while only 16 percent of large-cap CEOs believed they could deliver on that timeline. That six-month deadline has now arrived, and companies face pressure to demonstrate value.

These pressures have created a dangerous temptation. Many executive teams are considering indiscriminately restricting AI use to control costs and balance quarterly budgets. The authors warn that this reactive whipsawing will backfire: it will slow growth and efficiency gains already captured, and it will discourage experimentation, leaving companies unprepared for the next generation of AI capabilities. Instead, they argue the path forward is disciplined governance. They offer five specific recommendations. First, stop asking how much you are spending on AI and start asking where AI investment creates durable competitive advantage and where it generates consumption without compounding value. Second, treat AI spend as a capital allocation decision, not an IT budget line: usage that drives new revenue or builds proprietary capabilities is a growth investment, while usage that automates low-value processes is an operating expense. Third, establish governance mechanisms that match workloads to the least expensive model capable of performing them reliably, since employees naturally gravitate toward the latest models even when older generations can produce the desired output. Fourth, reshape incentives to reward efficient AI use—achieving better business outcomes with the appropriate level of consumption—rather than rewarding maximum usage or penalizing with blunt caps. Fifth, distribute AI governance throughout the organization rather than delegating it to a single role; managers across functions need to be accountable for sustainable adoption within their teams.

The authors frame this as the beginning of a second phase of AI adoption, moving from rapid competency-building to sustainable value extraction. They contend that token-based usage pricing has made AI costs highly visible, surfacing a broader governance problem. The companies that will win are not those that use AI the most or spend the least, but those that govern it best and consistently convert AI consumption into lasting economic advantage. The window to build this governance infrastructure is open but will not remain open indefinitely, and getting the balance wrong may soon be existential.

The article identifies a fundamental inflection point in corporate AI adoption: the shift from proof-of-concept and rapid deployment to disciplined, sustainable economics. This transition was triggered by two colliding realities. First, the financial model underpinning early AI enthusiasm—swapping human labor for model inference costs—proved less straightforward than executives anticipated. Organizations that pursued large-scale layoffs in anticipation of AI productivity gains discovered that they had eliminated the domain expertise required to integrate, refine, and operationalize AI systems effectively. Ford's high-profile rehiring of engineers exemplifies this cost; the company must now pay to rebuild the talent it shed. Second, a major investor-management misalignment has come due: Teneo's survey found 53 percent of investors expected AI return on investment within six months, while only 16 percent of large-cap CEOs believed that timeline was realistic. That deadline has now arrived, and executives face pressure to justify continued spending without clear returns.

The authors argue that the real competition will not be won by companies that spend the most or use AI most aggressively, but by those that govern it most effectively. Token-based consumption metrics have made AI spending highly visible and measurable—a double-edged sword that forces accountability but can also trap executives into reactive budget-balancing. The risk they highlight is that indiscriminate cost-cutting will backfire, discouraging experimentation and leaving companies unprepared for the next phase of AI capability. Instead, they propose a structural reorientation: treat AI spend as a capital allocation decision (with revenue-generating uses as growth investments and process automation as operating expenses), establish technical governance to route workloads to the least expensive capable model, and restructure incentives to reward efficiency per token rather than sheer usage. The implication is that the companies that master this balance—disciplined spending without stifling innovation—will accumulate durable competitive advantage, while those that fail at this balance face existential risk.

No comments yet. Be the first to share your thoughts!