Senior leaders embrace vibe-coding

When it comes to software, DIY might be the new “buy.”

Companies view vibe-coding as a way to reduce their reliance on traditional SaaS companies, EY research found. In the latest iteration of the Big Four firm’s US AI Pulse Survey, more than three-quarters (76%) of senior executives whose companies are investing in AI said that “off-the-shelf” software was insufficient for their needs. What’s more, 91% said that it’s “critical” for organizations to use AI to develop their own software. Almost as many (87%) either have programs in place to help employees build software with AI, or are piloting such programs.

The survey polled 534 US “decision-makers” (people with SVP roles or higher) from a wide range of industries between April 24 and May 17. Almost all (99%) came from organizations that have invested in AI, according to an EY representative.

In-house bumps. Vibe-coding is hardly frictionless. Nearly three-quarters (72%) of senior leaders said their organizations are “facing challenges” with AI-built tech. They cited a long list of potential problems associated with building software in-house, including the rise of shadow IT (34% said this was a barrier); concerns about regulation (33%) and cybersecurity (32%); and at 31% each, the higher initial cost involved, the lack of in-house talent, and the reliability and accuracy of AI. Only 8% said they experienced no barriers to developing their own software.

Almost all (98%) respondents whose organizations used tools that require tokens said they were rethinking their AI strategy due to cost. About a third (32%) of senior leaders were reconsidering which personnel would have access to AI.

Overall, AI may prove to be worth the money. Virtually all respondents (98%) saw positive ROI from the technology. And more than a third (37%) of the respondents who used tokens said they were considering expanding their use of AI, while only 15% were thinking of reducing it.

But the hype of the past few years may have cooled. In 2024, from 15%–18% of surveyed executives predicted that they’d allocate 50% or more of their budget to AI in 2025. In reality, only 4% ended up doing so.