Hello, dear readers!

This is our weekly brief on remarkable AI topics, so you can keep up without getting lost in the hype.

Today’s focus — most companies still aren’t seeing returns from AI. Despite years of investment and nonstop headlines, over half of executives report little to no tangible benefit — and only a small minority see gains on both revenue and cost. If AI is already everywhere, why isn’t it paying off?

Also in this week’s edition:

Most companies still aren’t seeing returns from AI

For all the talk of transformation, the reality inside most companies looks far less dramatic. According to PwC’s recent global CEO survey, more than half of executives say their AI investments have yet to deliver meaningful financial results. Only around 12% report seeing benefits on both revenue growth and cost reduction — the combination that actually moves the needle.

Notably, PwC itself continues to push companies to keep investing. As the firm’s CEO put it: “In periods of rapid change, the instinct to slow down is understandable — but it’s also risky. The companies that succeed will be those willing to make bold decisions and invest with conviction in the capabilities that matter most.”

And so far, spending isn’t slowing down. Companies continue to pour money into AI, driven by competitive pressure and fear of falling behind. No one wants to be the one who missed the shift — even if it’s not yet clear where the value will come from.

The divide is becoming harder to ignore. A small group of companies appears to be capturing real benefits, while many others are still searching for them. That raises a broader question about what AI actually is at this stage — a foundational technology that will take time to pay off, or more of a niche tool with limited impact in specific sectors.

If the gap between expectation and reality persists, the conversation may start to shift. Not toward abandoning AI, but toward a more sober question: what exactly is worth building — and what isn’t?

Allbirds drops shoes for AI

In one of the more surreal headlines this week, Allbirds — best known for its minimalist wool sneakers — announced a pivot toward AI.

The market reaction was immediate. The company’s stock briefly spiked by as much as ~600% before pulling back and settling closer to +300% — still an extraordinary move for a struggling footwear brand. For a business that makes shoes, becoming an “AI company” turned out to be a powerful narrative — at least in the short term.

Whether there’s substance behind it is another question. Becoming a real AI player takes years of investment, talent, and infrastructure. Calling yourself one takes a single announcement. For now, the latter seems to be doing most of the work.

Claude 4.7 arrives — and the benchmarks seem great

Anthropic has introduced Claude Opus 4.7, the latest iteration of its flagship model — and, as expected, the numbers are strong. The company reports improvements across reasoning, coding, and long-context tasks, positioning the model as a step forward in capability.

As always, though, benchmarks only tell part of the story. They offer a controlled view of performance, often optimized for specific tasks, and don’t always translate cleanly into real-world use. The broader question — how these gains show up in everyday workflows — remains harder to answer.

It’s also worth noting that some of the most talked-about capabilities, including the previously teased Mythos system, are still in preview. The direction is clear, and progress is real — but the gap between what models can do in testing and what they reliably deliver in practice hasn’t disappeared.

Thanks for reading AIport. Until next Monday — by then, AI will almost certainly get better, and harder to justify at the same time.