Workday Built Its Business Managing Humans. Its AI Agents Are Now Automating Them.
Workday's stock dropped 43 percent in the first four months of 2026. The S&P 500 gained 9 percent in the same period. The consensus bet was simple: AI would eat the human resources software market, and Workday — a company whose entire business is managing workforces at scale — would be first on the menu.
Q1 earnings suggest the market was wrong. Not slightly wrong. Catastrophically wrong, in the direction that punishes short-sellers and humbles the consensus.
Workday reported adjusted earnings of $2.66 per share against a $2.51 consensus estimate. More importantly, annual recurring revenue from agentic AI solutions is approaching $500 million, new annual contract value from those products grew more than 200 percent year-over-year, and the number of customers running at least one Workday-built AI agent more than doubled quarter-over-quarter to over 4,000. Expansion deals that included AI ran more than 50 percent larger on average than deals without AI, meaning existing customers are buying more — not replacing what they have.
On the earnings call, Bhusri argued that the 150th feature in HR or finance is not going to move the needle for their business — the next agentic application will. That was Aneel Bhusri, Workday's returning co-founder, who replaced Carl Eschenbach as CEO during the quarter — a leadership change that signals how seriously the company is treating the agent transition.
The bears had a coherent theory. If AI agents handle resume screening, interview scheduling, onboarding, and benefits administration, the logical outcome is that companies need fewer HR people. Fewer HR people means fewer seats to manage. Workday's per-seat pricing model — where it charges per employee covered — would then face systematic contraction. The stock collapsed accordingly.
What the theory missed is how agent-driven volume creates the exception workload that sustains human roles. AI agents still need to be configured, monitored, governed, and corrected. They handle the routine; humans handle what the routine generates. The agents that screen resumes also generate disputes when candidates get rejected. The agents that onboard new hires also produce compliance questions when roles span multiple jurisdictions. More agents processing more transactions does not reduce the exception layer — it expands it, because the agents' own outputs require human review, escalation handling, and edge-case resolution. Workday's platform processed approximately 1.4 trillion annual transactions across 80 million contracted users in Q1. That transaction volume is not shrinking as AI agents enter the workflow — it is growing, because the agents are processing more of the routine work that humans previously handled manually, and humans are now handling the higher-friction cases that the agents escalate.
The implication for Workday's pricing model is direct: more automation generates more throughput, more throughput generates more exceptions, and more exceptions require more human oversight. The per-seat model does not contract under agentic deployment — it potentially expands, because the same workforce now manages a larger volume of automated activity and the corresponding exception layer. That is the dynamic the stock re-rating reflects, and it is also the dynamic the bear thesis did not fully price in.
Contract Intelligence, Workday's AI tool for reviewing vendor and employment agreements, analyzed more than 1.1 million contracts in Q1, up 53 percent from the prior quarter. Recruiting Agent supported 14 million hiring processes, up 44 percent year-over-year. More transactions. More agreements reviewed. More hiring processes managed. The agents are not displacing the platform's workload — they are generating enough throughput that the platform's relevance expands.
The financial picture reflects this. Total Q1 revenue was $2.542 billion, up 13.5 percent year-over-year. Subscription revenue hit $2.354 billion, up 14.3 percent. Gross revenue retention held at 97 percent — meaning Workday is keeping essentially all of its existing customers and growing them. The company called it the best first quarter of new ACV growth in five years.
There is a ceiling, and it is worth naming directly. Agentic AI solutions are still under 5 percent of Workday's total revenue — the $500 million ARR figure sits inside a $10 billion annualized revenue base, and the company's overall growth rate of 13.5 percent is still the rate of a mature enterprise software company, not a platform in acceleration. Whether Workday's agent positioning converts to durable pricing power or simply delays the competitive pressure is the question the Q2 numbers will answer.
The open question — whether Workday's agent platform is open to third-party builders or locked to first-party agents — is one the company has not resolved publicly. If it opens, Workday becomes infrastructure for a broader agent ecosystem and the installed-base argument strengthens. If it stays first-party only, the agent layer is another feature inside a closed platform, and the distribution advantage is narrower than the bears assume.
The bears had a coherent theory, and the numbers do not yet fully disprove it.
What to watch next: whether Workday's 20 AI agents in general availability or early access can sustain the new-ACV growth rate through a full fiscal year. Bhusri's bet is that the agentic transition will take multiple years to play out and that Workday's installed base of 80 million users gives it a distribution advantage no pure-play AI vendor can match. The stock has re-rated sharply since the earnings report. Whether the multiple holds depends on whether Q2 and Q3 show the same pattern or whether the AI efficiency gains eventually compress the seat-count math the bears were originally betting on.