Q2 earnings season became an AI disclosure exercise
Across Q2 2026 earnings calls, a distinct pattern emerged among major retailers: nearly every one found a way to fold an AI initiative into its results narrative, regardless of whether the AI work was central to the quarter's actual performance. Gap disclosed capital expenditure of roughly $650 million for the year, spanning new stores, remodels, technology, and supply chain investment, with AI-related work sitting inside that broader number rather than reported as its own line. Dollar General, meanwhile, posted net sales growth of 5.2% year over year to $11.3 billion, a strong quarter delivered while the company continues rolling out AI-driven replenishment tools.
Ulta Beauty and Kohl's took a more explicitly customer-facing angle. Ulta CEO Kecia Steelman told investors, "We're in the early stages of applying AI across key corporate uses to enhance how we work, improve productivity and drive greater efficiency," while Kohl's CEO Michael Bender said, "We see significant opportunities to expand AI-assisted discovery, gifting and purchase confidence over time." Both statements are notable for what they do not claim: neither executive attached a hard revenue or margin figure to their AI programs specifically.
A Bain partner names the pressure driving the disclosures
Aaron Cheris, Bain & Co.'s global head of retail practice, offered the most candid framing of the quarter when he told reporters, "Everybody and their sister has to have an answer to the question of 'what are you doing with AI' for their board, for their investors." That is a materially different claim than saying AI is delivering results. It is a claim about a governance and investor-relations obligation that now exists independent of proven return, and it explains why so many earnings calls, across categories as different as beauty, department stores, and discount grocery, suddenly sound alike.
This matters because it flips the usual sequence of enterprise technology adoption. In the traditional version, a company proves a return internally, then discloses it to investors as validation of a decision already made. What Cheris is describing puts the disclosure obligation first, driven by board and investor pressure, with the proof of return expected to catch up afterward. That sequencing can still work out fine, but it carries a different risk profile than the one most capital allocation frameworks were originally built to evaluate.
The spend is real even when the framing is defensive
It would be a mistake to read Cheris's comment as evidence that the underlying spend is fake or purely cosmetic. Gap's $650 million capital expenditure figure and Dollar General's continued rollout of AI-driven replenishment tooling represent genuine capital and operating commitments, not slide-deck language. The pattern is less about companies inventing AI work to satisfy a board question and more about companies bundling real technology investment into a narrative that also happens to satisfy that question.
That bundling has a practical consequence for anyone trying to benchmark peer spending. When AI investment sits inside a broader capital expenditure or supply chain figure rather than being broken out, it becomes difficult to isolate what a retailer is actually spending on AI specifically versus general technology modernization. Boards asking "what are you doing with AI" are often getting an answer that blends genuinely new AI work with technology spend that would have happened anyway, relabeled for the moment.
Customer-facing AI is still described as early
What stands out across the Ulta and Kohl's commentary, and echoes what Walmart and Home Depot have said in prior quarters about Sparky and Magic Apron, is that even companies actively promoting shopping assistants describe the effort as early-stage internally. Steelman's phrase, "early stages," and Bender's phrase, "opportunities... over time," are both hedged language for public disclosures, the kind executives use when they want to show momentum without overpromising results they cannot yet back with hard numbers.
That hedge is worth taking seriously rather than dismissing as corporate caution. It suggests the retailers furthest along in customer-facing AI, the ones with named assistants and public usage statistics, are still the exception rather than the rule. Most large retailers are earlier in the process than their earnings-call language implies, which is a useful corrective for any technology leader benchmarking their own AI roadmap against what competitors claim to be doing.
Supply chain AI is getting less airtime than shopping assistants, and more budget
The consumer-facing AI assistants, Ulta's tools, Kohl's discovery features, Walmart's Sparky, Home Depot's Magic Apron, get most of the press coverage because they are visible and demoable. But the capital figures tell a different story about where the money is actually going. Gap's $650 million spans supply chain investment alongside stores and technology, and Dollar General's AI push is explicitly in replenishment, the unglamorous back-office function that determines whether the right product is on the right shelf.
That split, visible customer-facing AI for the board and investor story, less visible operational AI for the actual margin impact, is consistent with how enterprise AI adoption tends to play out once the initial hype phase passes. Supply chain and inventory AI has a clearer, more measurable path to return than a shopping assistant does, because inventory accuracy and replenishment timing translate directly into markdown avoidance and stockout reduction, figures a retailer's finance team already tracks closely.
What this means for how you frame your own AI budget
If your board is asking the same question Cheris describes, the retail earnings season offers a useful template for answering it honestly rather than defensively. Separate the disclosure you owe your board from the operational bets you are actually making, and be willing to say some of your AI work is early-stage rather than overselling a pilot as a platform. Ulta and Kohl's did this in public and it did not read as weakness, it read as credible.
The bigger structural takeaway is to prioritize the AI investment that improves a metric your finance team already tracks, inventory accuracy, replenishment timing, markdown rates, over the AI investment that is easiest to demo to a board. Gap and Dollar General's capital allocation suggests the more disciplined operators are doing exactly that, letting the customer-facing assistant carry the narrative while the real budget goes toward the supply chain work that is harder to see and easier to measure.
