What I Couldn't Say In HR Executive

The column got traction. Here's the unfiltered version.

My HR Executive column on what Anthropic’s pricing shift means for CHROs was published this week. The response was faster and louder than I expected, not from vendors, not from investors, but from HR leaders who recognized something in the argument that their own vendors haven’t told them yet.

So here’s the version I couldn’t quite publish in a trade column.

The vendor renewal conversation is already being shaped without you

I named names in the HR Executive piece, and I’ll name them again here: Workday, SAP, Oracle, and ADP are public companies. They have shareholders, analysts, and quarterly earnings calls to manage. The pricing shift happening at the frontier AI layer, the Anthropic move, the OpenAI move that’s coming, the Google move after that, will flow through their P&Ls and into your contracts. Not on your timeline. On theirs.

“Private vendors have more room to maneuver. They can absorb margin pressure longer, structure deals more creatively, and make pricing decisions without a quarterly earnings call forcing the issue. That doesn’t make them better. It makes them differently motivated.”

Motivation matters when you’re negotiating a multi-year contract in a rapidly shifting pricing environment. Most CHROs aren’t walking into that conversation with this framing. They should be.

The a16z piece that every CHRO should read this week

A16z dropped “Workday’s Last Workday” this week. If you’ve been following my Substack, you already know I’ve been in this debate. Last week, I published my own take on the a16z piece and the five responses it generated, where the smart observers are right, where they’re falling short, and what nearly a decade of WorkTech funding data actually shows about who is positioned to win.

For the purposes of this piece, here’s the part of that debate that matters most to CHROs right now:

The argument is architectural: Workday cannot become AI-native without starting over, and a public, installed-base company cannot do that. Every Illuminate feature is an additive overlay on the same forms-and-approvals engine that’s been in place for 15 years. Flex Credits exist because both sides need them; your CIO and CFO need “AI investment” as a line item, and Workday needs AI ARR on their earnings call. The question of which agent actually runs which workflow gets deferred until the product ships something useful. Both sides leave the room having checked their KPI box. The underlying architecture question goes unanswered.

The cost substitution nobody is talking about

Here’s the line from the HR Executive piece that got the most reaction:

“Agents doing more work does not automatically mean fewer people and lower costs if token consumption scales faster than labor savings. What looks like a headcount reduction opportunity may actually be a cost substitution, trading predictable salary expense for variable, difficult-to-forecast compute expense.”

I’ve been in this industry long enough to watch AI business cases get built on assumptions that don’t survive contact with the finance team. This one is particularly vulnerable because it looks airtight on a slide. Headcount goes down, AI spend goes up, net savings are positive. The problem is the variable cost exposure hiding in the denominator, token consumption that scales with every agentic workflow you add, every employee who brings their own AI agent to work, every platform integration your vendor builds on your behalf inside a contract you signed before any of this existed.

Your CFO will find it. Better that you find it first.

The five questions worth asking before your next renewal

I laid out five strategic imperatives in the HR Executive column. Here they are as questions instead; the ones I’d be asking if I were sitting in the CHRO seat right now:

Do I know my organization’s AI consumption footprint across my HR stack? If not, Finance will figure it out for me and make decisions about my tools based on cost rather than value.

Does my current vendor contract reflect consumption-based economics, or was it written for a per-seat world? If the latter, what happens when usage scales beyond the original estimate?

Is HR in the enterprise AI strategy discussion? Capacity planning, headcount strategy, role redesign, governance- these are HR questions. If I’m not in that conversation, someone else is answering them.

Have I piloted anything AI-native, or am I relying entirely on my incumbent platform’s roadmap? One scoped project in an adjacent budget line is worth more than a year of analyst briefings.

Can my vendor prove the outcome? Who owns the data that measures it? What happens to pricing if we don’t hit the targets?

The vendors who can answer those last questions clearly are the partners worth betting on. The ones who can’t are selling a talk track.

The bottom line

I’ve written two pieces on this shift now: the vendor side in WorkTech, the buyer side in HR Executive. The thread connecting them is the same: the subsidized era of enterprise AI is ending, the cost of compute is real, and the gap between vendors who are operationally ready to deliver on their pricing promises and those who aren’t is wider than most buyers realize.

The HR leaders who get ahead of this will negotiate from a position of strength. The ones who don’t will find out at renewal time, when the leverage is gone.

The bill is coming. Read the full HR Executive column before it arrives.

HR Executive column | WorkTech vendor piece