A new study by PwC, published on April 13, 2026, reveals a stark and widening divide in corporate AI investment returns. The 2026 AI Performance Study, based on interviews with 1,217 senior executives across 25 sectors, found that 74% of all AI economic value is captured by just 20% of organizations. This aligns with earlier MIT research from August 2025 showing that 95% of enterprises reported zero return on generative AI pilot projects.
PwC's earlier January CEO survey of 4,454 executives across 95 countries found that 56% saw neither higher revenue nor lower costs from AI over the prior year, with only 12% achieving both benefits simultaneously. According to separate PwC analysis, companies applying AI to products, services, and customer experiences achieved nearly four percentage points higher profit margins than those that did not.
PwC global chairman Mohamed Kande stated, "a small group of companies are already turning AI into measurable financial returns, while many others are still struggling to move beyond pilots." The firm frames the divide as structural rather than temporary, attributing success to strong AI foundations including a supportive technology environment, a defined roadmap, formalized risk processes, and an adoption-friendly culture.
Simultaneously, Stanford HAI's 2026 AI Index, released the same week, documents a growing transparency crisis. The report finds that the most powerful AI models are the least transparent about their training data and benchmark performance. "A lot of companies are not releasing how their models do in certain benchmarks, particularly the responsible-AI benchmarks," the report states, citing growing opacity as accountability becomes more critical.
The benchmark system itself is failing, with some popular tests carrying a 42% error rate, while others can be gamed by models trained on the test data itself. This means strong scores do not reliably indicate stronger or safer real-world models. The report notes that for complex use cases like AI agents and robots, benchmarks barely exist, meaning the most consequential applications are deployed with almost no standardized external validation.
US trust in the government to regulate AI sits at just 31%, the lowest of any country surveyed. Globally, the EU is trusted more than either the US or China, a finding that reflects the full enforcement of the EU AI Act in January 2026 and the absence of a comparable federal framework in the US.
Gartner describes the current moment as AI's "Trough of Disillusionment," where experiments fail to deliver and the market separates into survivors and casualties. PwC warns that the gap between AI leaders and followers will widen quickly for those that don't act, making 2026 the year the divide becomes durable rather than correctable.
