AI has captured boardroom and investor attention at a scale rarely seen outside of major platform shifts. That visibility has driven unprecedented investment and experimentation, but it has also fueled skepticism. The debate is no longer about whether AI works, but whether expectations have outpaced the realities of enterprise adoption.
Those arguing that AI is overhyped typically point to capital markets behavior, infrastructure spending, and unclear near-term returns.
Several investment and research firms have highlighted the concentration of market value in a small number of AI-exposed companies, paired with aggressive forward-looking assumptions. Goldman Sachs has compared elements of today’s AI investment cycle to the early internet era, noting heavy infrastructure investment ahead of proven monetization and uncertainty around long-term winners.
A similar caution appears in commentary from Sequoia Capital , which raised concerns about whether projected enterprise spending on AI can realistically offset the cost of building and operating large-scale model infrastructure. Their analysis emphasized that productivity gains must scale materially for current investment levels to be sustained.
In addition, Gartner has warned that generative AI has entered the “peak of inflated expectations” phase of its hype cycle. Gartner’s research suggests that many organizations adopt AI based on perceived competitive pressure rather than validated business cases.
Together, these perspectives frame AI not as fiction, but as a technology whose economic impact may lag its narrative.
In contrast, others argue that labeling AI as overhyped misunderstands both the maturity curve and where value is actually being created.
Research from McKinsey & Company shows that organizations considered AI leaders are already capturing measurable gains in productivity, forecasting accuracy, and operational efficiency, particularly when AI is embedded into core processes rather than deployed as standalone tools.
Academic research from Stanford University , through its annual AI Index, shows steady improvements in model performance, cost efficiency, and accessibility. While adoption outcomes vary widely, the underlying technical progress has been consistent.
Market observers also note that early AI attention has focused heavily on visible interfaces, such as chatbots, while less attention is paid to embedded AI systems that quietly optimize planning, scheduling, forecasting, and decision support. This mirrors previous technology cycles where enterprise value emerged after early hype faded.
From this perspective, today’s tension reflects a shift from experimentation to execution, not a collapse in capability.
Enterprise adoption data helps explain why both sides of the debate feel justified.
According to McKinsey & Company, nearly 90 percent of organizations report using AI in at least one business function. However, only a small fraction report that AI meaningfully impacts enterprise-wide performance metrics. Most deployments remain localized, experimental, or dependent on individual users rather than embedded into operating models.
Research from MIT, as reported by Fortune, suggests that the majority of generative AI initiatives stall after pilot phases. As many as 95 percent fail to demonstrate sustained business value, not because the technology is ineffective, but because it is not aligned with workflows, data foundations, or organizational ownership.
Additional surveys from Deloitte reinforce this finding, noting a persistent gap between executive enthusiasm and operational readiness, particularly in regulated and asset-intensive industries.
For energy and commodity organizations, the AI hype debate is not theoretical. It directly affects capital allocation, technology roadmaps, and risk exposure.
These industries operate in environments where technology must be reliable, auditable, and clearly tied to operational and financial outcomes. Research from the International Energy Agency shows that AI delivers the most value in energy when applied to forecasting, system optimization, asset performance, and risk management, not generic productivity experimentation.
In this context, caution is not resistance. It is a filter. Energy organizations that succeed with AI tend to anchor initiatives to specific decisions, integrate AI tightly with trading and operational systems, and establish governance early. Those that do not often accumulate pilots without impact.
The real risk for energy companies is not missing the AI wave. It is adopting AI without discipline.
Ready to apply AI in practical, measurable ways?
Start a conversation with our team.
It’s important to know who you’re working with, we get it. Meet the VG team and explore our backgrounds that bring you value.
