Two-thirds of IT leaders see AI results, but few see urgency

The contrast captures the gap between AI’s growing presence in companies and the harder question of whether its returns justify the investment. That question matters well beyond individual budgets: the industry’s data-center buildout depends on companies continuing to spend, ideally at rising prices.

Results are not the same as returns

Azhar, a British technology entrepreneur and founder of the research group Exponential View, shared the exchange on a podcast with Nicholas Thompson of The Atlantic. He had spoken with about 160 IT vice presidents in Las Vegas. Two-thirds could name measurable AI results, more than he expected; only about eight judged those results important enough to interrupt a CEO’s summer vacation.

Azhar said companies are making progress, but slowly. Whether that pace can support current investment levels remains unclear. Some executives say that after early successes, boards are setting more ambitious goals. Azhar also heard from CEOs in slower markets such as Italy that confidence in AI is growing and budgets are increasing, despite mistakes along the way.

A company can report useful results without showing that AI will generate enough money to sustain the industry’s infrastructure spending. The Las Vegas answers suggest that measurable gains are already common; the much smaller number willing to treat them as urgent points to a different standard of proof.

The pressure to tell a success story

Many companies are moving from expensive frontier models to open alternatives. That could increase AI use while weakening the economics behind the data-center expansion: adoption alone does not ensure enough money flows back to fund new capacity.

Azhar called the downside case the “bearish version of the story.” A Boston Consulting Group survey found that about 70 percent of CEOs worldwide consider AI success important to their professional reputations. That creates a reason to describe results more optimistically than they are.

I think the more telling signal is not that two-thirds can name results, but that only about eight people saw enough value to interrupt a CEO’s vacation. That is a small sample, not an economy-wide verdict. Still, it separates progress from the kind of payoff that can carry a large investment cycle.

Azhar has no simple answer on whether AI is in a bubble; he described the situation as “finely balanced.” The missing evidence is not another claim that AI is useful, but proof that its returns can keep pace with the infrastructure being built to support it.