Memorandum
- From
- Derek Weston via Fast Company
- Date
- Filed
- Business·5 min to read
- Re
Corporate AI adoption lags behind hype but is steadily accelerating
ReCorporate AI adoption lags behind hype but is steadily accelerating
Less than a quarter of S&P 500 companies have deeply integrated AI into their operations, according to a new analysis of corporate filings. The technology sector leads, while non-tech firms, which make up 90% of the U.S. economy, are moving more cautiously but show growing momentum.
Corporate America is adopting artificial intelligence at a pace far slower than the relentless stream of headlines suggests, yet the trajectory is unmistakably upward. A new analysis of federal financial filings from S&P 500 companies reveals that fewer than one in four firms had AI deeply integrated into their business processes or used it in producing goods and delivering services by the end of last year. The findings challenge both the narrative of overnight transformation and the counter-narrative that AI is entirely overhyped.
The technology sector is clearly in the fast lane, accounting for two-thirds of extensive AI integration and use. Outside of tech, fewer than two dozen S&P 500 firms — including Moderna, Mastercard, Bank of New York Mellon, and GE Healthcare — have achieved what the analysis defines as full deployment, meaning AI is a core component of strategy and financial performance, embedded across business functions. This disparity matters because non-technology firms make up roughly 90% of the U.S. economy, and it is among these enterprises that the largest untapped value lies.
Several factors are slowing the pace. For AI to meaningfully change how work gets done, it must perform tasks with access to the right information and do so cost-effectively. In many cases, running AI models at the required level of precision remains too expensive. The analysis notes that the tasks AI can theoretically automate still far exceed those it actually automates today. However, capability is improving rapidly. A task like creating a 10- to 12-slide presentation for a quarterly customer review, which once took three to four hours, can now be attempted by large language models with a 50% success rate from two years ago, improving to 65% a year later. If current trends hold, models could complete most text-related tasks with good-enough quality at success rates of 80% to 95% by 2029.
More than half of the S&P 500 currently have AI pilot projects underway. Many will fail, but the lessons learned will be copied by others, portending significant gains in productivity, efficiency, revenue, and profits. The analysis points to practical applications that go far beyond using ChatGPT to draft a job description. A supermarket chain with 500 stores, each handling 50,000 stock-keeping units, faces the challenge of forecasting demand for 25 million items every two weeks. AI deployed to project demand more accurately and share that information with vendors could reduce wasteful spending in a low-margin industry. Similarly, in manufacturing and construction, AI can help project managers identify whether work is on time and on budget, and pinpoint where and why misses occur.
Beyond improving existing processes, AI is expected to be critical for creating new products and services, from accelerating drug development at pharmaceutical companies to engineering drought-resistant seeds and enabling more innovative vehicle designs. None of this is guaranteed. Companies must reorganize workflows and foster cultures that encourage employees to embrace AI-augmented roles. The analysis acknowledges that AI will lead to layoffs, in some cases on a large scale, but expresses optimism that most businesses will move beyond simply substituting technology for workers. By combining human and machine efforts, leaders can devise new products, introduce new processes, and create better tools, which should translate into new hiring.
A key insight from the analysis is the role of partial automation. Achieving good accuracy on a task may be relatively inexpensive, but pushing from good to near-perfect accuracy can be significantly more expensive for both AI builders and users. When the marginal cost of further accuracy exceeds the marginal labor saving, firms are better off stopping short of full automation. This economic reality suggests a future where AI handles substantial portions of work while humans complete the remainder, a model that could define the next phase of corporate adoption.
