The Model Was Never the Moat — Enterprise AI 2026

Koko: Forty-two percent of enterprise AI deployments found the model interchangeable. Not better than the competition's model. Not worse. Interchangeable. That one number basically blows up two years of breathless coverage about which foundation model wins.

Max: It's a brutal finding. And it comes out of the Stanford Digital Economy Lab, so it's not some vendor trying to sell you something adjacent. The model — the thing everyone's been in a bidding war over — is already a commodity input.

Koko: And the piece synthesizes twenty-two studies from 2026 — BCG, McKinsey, Bain, Stanford HAI, Microsoft, Accenture, PwC, the full roster — and they all point at the same conclusion. The frontier model is not the moat. The question is: what is?

Max: Which is exactly the right question, and also the one most organizations are not funding.

Koko: So let's make the thesis really crisp, because the piece earns it. The argument is not just that models are commoditizing. It's that the advantage has moved to the organizational layer — and almost nobody is investing there.

Max: Right, and three facts stack on top of each other in a way that's almost uncomfortable. About eighty-eight percent of organizations are using AI in at least one function. Agent deployment — the next step up — is still in the single digits. And only a thin slice report sustained, enterprise-wide impact.

Koko: The piece calls that gap — between those three numbers — the entire twenty twenty-six enterprise AI story. And I think that's right.

Max: It's the difference between having a gym membership and being fit. The access question is basically solved. The conversion question is where everyone is stuck.

Koko: And the money is absolutely moving. BCG finds AI is the top investment priority for the next two years. Budgets have roughly doubled as a share of revenue — from about zero point eight to one point seven percent year over year. And the overwhelming majority intend to keep increasing spend.

Max: Even though a majority of CEOs — Bain puts it at more than eighty percent — are already dissatisfied with the results. So you have surging spend against unproven return, and the piece is very honest that that tension is the defining story of the year.

Koko: Let's go into the evidence, because the sourcing here is genuinely rigorous — the piece flags that four claims from an earlier draft were struck or re-anchored. So the numbers that survived are worth sitting with.

Max: Start with the organizational-versus-technical split, because it's the one that should change how executives allocate budget. Microsoft's Work Trend Index finds that organizational factors drive roughly twice the AI impact of individual skill — a sixty-seven to thirty-two percent split.

Koko: And Stanford independently finds that seventy-seven percent of the hardest deployment challenges are organizational, not technical. Two different methodologies, same diagnosis.

Max: And yet — only fourteen percent of enterprises have a documented AI strategy with clear goals. The HFS and Altimetrik finding. The rest are defaulting to cost-cutting. Which is not a strategy.

Koko: The high-performer number from McKinsey is the one that really concentrates the mind. Only about six percent of organizations are what they call AI high performers — the slice attributing more than five percent of EBIT to AI. Six percent.

Max: And McKinsey ties that directly to workflow redesign and operating-model change — not model choice. So the winners have done the hard, unsexy organizational work.

Koko: There's also a labor market finding that I think will hit differently depending on who's listening. Anthropic's data shows hiring for twenty-two to twenty-five year olds running about fourteen percent slower. And Stanford's 'Canaries in the Coal Mine' research logs a roughly twenty percent drop in employment for that exact cohort in software development specifically.

Max: Which is not the broad displacement story people were arguing about. It's very targeted. Entry-level coding. And it's already in the data.

Koko: While at the same time, PwC finds that AI-exposed firms are growing headcount and seniority faster. So it's bifurcating — not collapsing.

Max: The boring answer is always 'it's complicated,' but in this case the data actually earns it.

Koko: Here's where I want to push on the piece a little, because it's honest enough to name where the data argues with itself — and that's the most interesting section.

Max: Go ahead.

Koko: Ninety-four percent intend to keep investing, per BCG. But more than eighty percent of CEOs are already dissatisfied with results, per Bain. That's not a rounding error. That's cognitive dissonance at scale. Why are organizations doubling down on something they admit isn't working yet?

Max: I'd argue it's fear of falling behind more than conviction about return. But the piece adds another contradiction that's almost funnier — seventy-two percent of programs are CEO-led. So it's not like there's no mandate from the top.

Koko: And yet organizational conditions drive twice the impact, and only fourteen percent have a clear strategy. So CEOs are leading initiatives into organizations that aren't ready to receive them.

Max: There's also the visibility problem. Budgets are climbing toward five percent of spend — Capgemini goes from three to five percent of budget — but only about twenty-six percent, per KPMG, have real-time visibility into what AI is actually costing them. You're accelerating and the speedometer is broken.

Koko: Which is where I'd actually push back a little on the framing. The piece implies the gap is mostly a failure of organizational ambition. But I wonder if some of this is rational patience — early-stage infrastructure spending that hasn't compounded yet.

Max: Maybe. But the CFO section puts that in tension pretty directly. Only about twelve percent have machine learning in FP&A forecasting at full scale. That's not early-stage patience. That's the finance function not eating its own cooking.

Koko: Fair. Physician, heal thyself.

Koko: Alright, so what does a smart executive actually do with this? Because the piece doesn't stop at diagnosis — it has a real point of view on the action.

Max: The KokoAI framing is blunt: spend that expires is a cost; spend that becomes an owned, auditable capability is equity. The goal is not to rent intelligence from a model provider. It's to manufacture governed, reusable assets that compound.

Koko: And they operationalize that with a specific set of things the CFO should be instrumenting — before scaling, not after. Total cost of AI ownership in real time. Impact attributed to specific workflows. An asset-reuse rate — are you reusing what you've built, or rebuilding it from scratch each time? And governed exposure — control coverage, human-in-the-loop thresholds, audit trails.

Max: The asset-reuse rate is the one I'd highlight, because it's the most honest signal of whether you're compounding or just spinning. If your AI spend keeps starting from zero, you're not building a moat — you're buying the same pick and shovel over and over.

Koko: And there's a trust dimension that the piece ties together from three independent sources — Forrester names it a 'trust tax,' McKinsey makes responsible-AI maturity a precondition for value, HFS shows only fourteen percent with a clear strategy. The recommendation is: price the trust tax instead of paying it blind.

Max: Treat it as a figure, not a feeling. Which is very CFO-coded advice, and I mean that as a compliment.

Koko: On the finance function specifically — and this is striking — only fifteen to twenty-five percent of CFOs have fully scaled AI in their own function. Of those who have scaled it, forty-one percent are satisfied. Of those still in pilot, only twenty-five percent are. The gap between scaling and piloting is sixteen percentage points of satisfaction. At some point you have to leave the airport.

Max: And ninety-five percent of CFOs say they'd pay a premium for solid AI capability, and ninety-two percent would shift labor budget toward it. The conviction is there. The follow-through is where it breaks down.

Koko: Here's the line from the piece that I keep coming back to: 'If twenty twenty-five was about access to AI, twenty twenty-six is about accountability for it.' That's the cleanest summary of where we are.

Max: And accountability, in practice, means one thing: the CFO owns the conversion of spend into a measurable, governed, compounding asset. Not the CTO. Not the head of transformation. The CFO. Because this is a capital allocation question now.

Koko: Twenty-two studies, one uncomfortable conclusion. The model was never the moat. The organization around it is. And most organizations are funding the commodity and starving the actual source of advantage.

Max: The forward question — and I genuinely don't know the answer — is whether the eighty-plus percent of CEOs who are dissatisfied right now are going to course-correct toward the organizational work, or whether they're going to keep throwing budget at better models and wonder why nothing changes.

Koko: That's the bet. And it's a big one. The full piece — with its filterable evidence ledger and every source verified to its publisher of record — is on Koko Knows. Worth reading before your next board conversation about AI spend.