Workday is arguably the most important and least loved product in enterprise software. More than 10,000 organizations run it, tens of millions of employees live inside it, it’s approaching $10 billion in annual revenue, and the public market values it at roughly $30 billion.
None of that success is because the product is beloved. HR admins spend their days on data entry, workarounds, and hand-holding: running reports across three screens, managing comp cycles in Excel because rolling them out in the system is too complicated, sitting on Zoom calls while business partners click through promotion workflows, waiting for IT to explain why an integration broke this week.
When customers renew at close to 100% every year, it’s usually read as a sign the product is delightful. In Workday’s case, it’s a sign of something else: leaving is close to impossible.
HCM is the last large enterprise software category without a serious AI-native challenger, and that’s about to change. A platform shift more transformational than the one that created Workday is now underway, and soon we’ll see Workday’s last workday.
The cloud shift that created Workday
Workday is itself the product of a platform shift. In 2005, after losing a hostile takeover fight to Oracle (a story worth its own post), the leaders of the then-dominant PeopleSoft, Dave Duffield and Aneel Bhusri, bet that the move from client-server to multi-tenant cloud would reset the HRIS category. They saw that Oracle and SAP would struggle to close the architectural gap, and within a decade every client-server HRIS was a legacy maintenance business. They also saw that large enterprises were turning away from massive upfront CapEx outlays and toward predictable annual OpEx subscriptions, moving off perpetual licenses and on-prem hardware onto subscription software.
Workday won the platform shift on two fronts at once: it was the only major HRIS architected for multi-tenant cloud from day one, and it was already priced for the subscription procurement model that enterprise buyers were converging on. Within a decade it became the default answer.
Why Workday has been so defensible
Every serious attempt to displace Workday at the enterprise layer has failed for the same reason: Workday’s moat isn’t really made of product. It’s made of everything around the product. This consists of:
- Deep technical and human wiring. Workday sits at the center of hundreds of integrations (payroll, benefits, ATS, expense, identity, finance, state tax), none of which migrate cleanly. Thousands of hours of muscle memory live inside each instance: the admin who’s run the same performance cycle four years in a row, the payroll manager who’s memorized the seventeen-step merit rollout. Adding a single new cost center can ripple through reporting, integrations, job architecture, comp grades, and benchmarking, and only works if every system and every person is updated in sequence.
- A proprietary configuration layer. Workday isn’t configured the way most enterprise systems are. Integrations are built in Workday Studio, a proprietary tool that takes 6 to 18 months and requires hands-on work from certified consultants. Reports run through Workday’s own BIRT implementation, calculated fields through Workday’s own expression syntax, and tenant configuration through business process and security frameworks unique to the platform. None of these skills have a market outside Workday, no open-source community, and no Stack Overflow, means both the developers and the customers who employ them are locked in by their own résumés.
- A services cartel. More than 10,500 certified consultants work on Workday implementations worldwide across Accenture, Deloitte, Kainos, PwC, KPMG, and 150+ smaller shops. Implementations take 6 to 18 months and cost $300K to $1M+, with implementation fees typically equal to 100% of annual software fees. That services economy is arguably a bigger asset than the product itself. It lobbies on Workday’s behalf, absorbs customer complaints, and gives the biggest firms in the world a vested interest in keeping Workday installed.
- Multi-year contracts. Workday locks customers into multi-year deals. Even a customer who would switch tomorrow is structurally forced to wait.
The combination is why Workday posts some of the best gross revenue retention in enterprise software, and why, for most large companies, it’s one of the stickiest products they own.
Why now
I can hear some people saying: “Sure, challengers have been trying to attack Workday for two decades. But it’s nearly impossible to break in.” And these people are right: the newer entrants have either gone after net-new startups as customers (a strategy pursued by companies like Rippling and Gusto) or competed orthogonally by picking up hard-to-serve markets, like Deel has done.
Meanwhile, Workday hasn’t been sitting idly while AI threatens it. It released more than 25 AI features in 2025 under the Illuminate brand, shipped more than a dozen agents, acquired Sana Labs and Pipedream, and rolled out a consumption-based Flex Credits pricing model that Accenture, Nike, and Merck have signed onto. AI ARR has crossed $400 million and is growing triple digits year over year.
While this seems like real commercial traction, Flex Credits and “AI ARR” are more of a procurement innovation than a deployment one. Signing onto Flex Credits isn’t the same as running core HR workflows through agents in production. Flex Credits exists because both sides need it. Every enterprise CIO and CFO has “AI investment” as a top-line KPI for 2026 and needs to show real spend against it; every legacy vendor has “AI revenue” as the headline question on every earnings call and needs to show real revenue against it. Flex Credits is the procurement structure that gives both sides what they need from the same dollar. The customer commits a credit pool that counts as AI spend in the budget; Workday books the commit as AI ARR; the question of which agent actually runs which workflow gets deferred to whenever Illuminate ships something useful. Both sides leave the room having delivered against their KPI, no one had to commit to a specific deployment, and they still got to go to Ruth’s Chris to celebrate the deal.
You don’t have to take our word for it — there are clear signs this is what’s happening with customers today. A large Workday services partner recently wrote that “most organizations have no idea these capabilities exist, let alone how to activate them.” Customers are already pushing back on paying incremental tokens on top of a subscription they already renewed, and the long-time Workday admins we’ve talked to describe Illuminate as the same manual admin work with a chat surface bolted on. Every Illuminate feature is an additive overlay on the same forms-and-approvals engine, which means AI ARR can grow triple digits while the underlying product stays exactly what it was.
But this moment is different: three things have changed that make Workday finally vulnerable.
First, enterprise IT is finally revisiting its core systems. Large enterprises are running AI-readiness reviews across the systems they thought were locked in for a decade — ITSM, ERP, HCM. The pace of change in the AI stack has turned legacy architectures into liabilities, and a company that wants to be forward-leaning on AI can’t do it on an HRIS built for 2005. When a CHRO and CIO ask “what’s the AI-native version of this?” and the answer from Workday is consumption credits on top of the same engine, that’s an opening.
Second, the tools to actually rebuild this now exist. The same motion is already running one tier up the enterprise stack: companies like Tessera are doing AI-native SAP migrations at Fortune 500 scale, a category an order of magnitude harder than HCM, where a single ECC-to-S/4HANA upgrade can cost $700M and take three years. HCM is the same shape of problem at a smaller surface area. Pair that with a forward-deployed services model owned by the vendor instead of Accenture, and the implementation layer stops being the moat it used to be.
Third, Workday structurally can’t close the gap from inside. The company is making three specific bets: that Illuminate is the agentic product customers want, that Sana is the new “front door for work,” and that its forthcoming Agent System of Record will be the governance layer for every agent in the enterprise. All three are layered on top of the same forms-and-approvals engine, which is very powerful, but its twenty-year-old, ossified legacy infrastructure that’s hard to configure and modify to accommodate changes a modern HR organization actually needs to make. Bolting AI on top of that engine doesn’t change what’s underneath.
The one real underlying asset, Workday’s trillion-transaction dataset, is thinner than it sounds; what actually matters at runtime is how data connects to workflows, permissions, and integrations, and every layer of that stack is now a liability. Workday can bolt AI on top of it, but it cannot become AI-native without starting over, and starting over is the one thing a public, installed-base company cannot do.
As we’ve discussed the F500 replacement cycle is about to open for the first time in two decades, thanks to a wave of AI-driven replatforming across enterprises. None of the next generation HR solutions were architected to be the F500 HRIS system-of-record. This is a unique opportunity to target the enterprise segment where Workday makes most of its money, and where every previous attempt to displace it has failed.
What an AI-native Workday should look like
We think the opportunity is to go head-to-head with Workday on Human Capital Management (HCM) and build an enterprise-grade AI-native HR system of record, designed for the F500 for the next twenty years.
The product we’d want to fund has six properties:
- Deploy in one month: coding agents collapse 12-month implementations into 1.
- Workbench-native: HR teams become analysts and builders, not ticket-writers.
- Agent-first: employees work with HR inside the tools they already use.
- Open: customer agents can read the HR data model directly.
- Secure and permissioned: intuitive per-agent access controls built into the system.
- Always-on compliance: monitoring agents track the regulatory surface in real time.
Deploy in one month. Implementation is Workday’s biggest weakness and the most significant reason enterprises don’t switch. A real enterprise Workday implementation requires specialists across 50-state US payroll tax, multi-country payroll across 60+ jurisdictions, ACA and SOX controls, benefits carrier quirks, union agreements, Workday Studio, BIRT reporting, and Extend configuration. No one person knows all of it, and most implementations run 12 to 18 months specifically because the knowledge is fragmented across a dozen specialists who have to be sequenced and coordinated. Coding agents collapse the fragmentation. A single agent can ingest the entire tenant (business process definitions, integration definitions, audit logs, payroll runs), reconstruct the rules in plain English, reconcile against live integration feeds, preserve the edge cases, and generate a configuration draft in days rather than quarters. Workday today is configurable within its allowed boxes. An AI-native HRIS should be customizable to the company’s actual policy, with coding agents doing the work a consulting firm used to quote at six figures.
A built-in workbench for HR teams. The best HR admins are product people at heart — they know which cross-system reports the CHRO actually needs, what a comp planning tool should look like, what an onboarding experience should feel like. Today none of that is one job. Pulling information across systems requires a data warehouse and a data team to map it. Building a real workflow or app requires a developer or a Workday Extend contract. The workbench collapses all of it into one agent-native surface: ask a cross-system question and get the answer, describe an app in plain English and have the platform generate a working one, propose a process change and see what it would do. HR teams we’ve talked to are already trying to build apps like this — a manager-onboarding flow that drafts the JD, assembles a 30-60-90 plan, and coordinates with IT on accounts and equipment.
Agent-first. In addition to having a portal, employees should interact with HR through the tools they already use. An employee traveling to Milwaukee should be able to ask in Slack who else from the company is within 50 miles and get the answer in the same thread. A manager approving a vacation request should see the employee’s full context (balance, recent time off, upcoming leave, team coverage) inline with the approval, instead of having to click through to a separate dashboard. Take something more involved: spinning up a new division. Today it’s weeks of work inside Workday: new cost centers, job architecture, headcount plans, benefits setup, payroll integration, approval workflows. In an AI-native system, an HR ops lead should be able to describe it in plain English (500 people in Austin and Dublin, reporting to this EVP, these job families, this comp band) and have the system map every dependency, flag the downstream changes, and generate the configuration and rollout plan in one pass. And the data should flow the other way: HR data should power agentic flows across the rest of the org, not stay trapped inside the HRIS.
Open. Connecting a new payroll vendor is a six-to-twelve-month custom integration through Workday Studio; adding a benefits carrier is a similar build; pulling data into a BI tool runs through a consulting firm. The practitioners we’ve talked to have stopped waiting. They’re building their own Claude MCPs to pull Workday data into the tools they actually work in, standing up custom apps to route approvals through Slack, and treating Workday as a read-only system of record behind an experience they built themselves. The HRIS these teams would build if they had the time is open by default: a data model the customer’s own agents can read directly, APIs that aren’t gated behind credit pools, and a connection layer that treats integrations as a first-class product. The ecosystem pull comes from shipping the best agent builder on top of the data, which is where the work gets done fastest.
Secure and permissioned at the agent layer. HR data is the most sensitive in the company (comp, performance, medical leave, PII), and agents acting on it need fine-grained access controls that are native to the system, not bolted on. A manager’s agent should see their team’s comp; an IC’s should not. An external recruiter agent should see open reqs; it should not see severance history. Getting per-agent permissioning right is the difference between AI that can actually touch production HR data and AI that’s blocked from it by policy, and it’s close to impossible to retrofit into an architecture that wasn’t built for it.
Always-on compliance. The regulatory surface around HR data (EU AI Act, GDPR, data residency) is expanding faster than any single admin can track. Inside Workday, staying current means a senior HR leader reading newsletters and hoping she didn’t miss something. An AI-native stack flips that: an always-on agent monitors for regulatory changes across jurisdictions, flags what needs to change in the tenant, and drafts the config update. That’s hard to retrofit into a 2005 architecture, and natural to build into a 2026 one.
How we’d start
Building all six will take time. Here’s how we would start.
Find a handful of F500 design partners running serious AI-readiness reviews of their HR stack. Lead with a mapping and eventual migration tool to surface the company- and geography-specific rules and edge cases that live inside the tenant, then start automating the human work that piles up around Workday (comp spreadsheets, performance hand-offs, ticket queues) before the full system-of-record swap is on the table.
The commercial architecture matters here. Workday’s multi-year contracts lock the HRIS line item for years at a time, but F500 HR organizations have adjacent budgets that aren’t locked: HR operations, HR technology, transformation, innovation, consulting. A scoped implementation project or an automation subscription can be sold cleanly into one of those, with a real SOW and real procurement, without fighting the Workday renewal on entry. By the time that renewal actually opens, the company is already inside the tenant and delivering value the CHRO can point to. At that point, the question stops being “try an unknown vendor” and becomes “expand the vendor we already trust into the budget we were going to spend anyway.”
It’s also important to consider Workday leveraging scope to crush startup competitors as most F500 Workday tenants are full-platform (HR, Finance, Payroll, Adaptive Planning). The CIO is not going to unstitch their entire tech stack to bet on an HR-only challenger that can’t solve their problems. The best answer is the playbook Workday itself ran against PeopleSoft: start narrow where the leverage is, partner into the adjacent scope where the customer already has a system, and build native where the strategic bet is. A new entrant can treat the customer’s existing Finance and Payroll instances as stable integrations to support on day one, replace the Workday HR modules the customer actually hates (performance, comp planning, reorg, natural-language reporting), and let the platform grow into the remaining scope over time. The sale leads with continuity: payroll keeps running, integrations don’t break, the renewal cycle doesn’t create a gap in how the company is run.
Each piece of this dissolves a layer of defense: agent-native workflows replace twenty years of muscle memory, natural-language configuration retires XpressO, a forward-deployed team routes around the services cartel, and the adjacent-budget wedge neutralizes the multi-year lock.
Don’t expect Workday to go quietly
Workday is already mobilizing. The company has cut more than 2,100 employees across two rounds of layoffs in the past fourteen months and co-founder Aneel Bhusri returned with an explicit mandate to pivot the company toward AI. A $30B company with 10,000+ customers and a services cartel that’s arguably bigger than the product itself will not roll over quietly.
Expect the full playbook. Aggressive bundling with Adaptive Planning, Payroll, and the Finance cloud to make the HRIS renewal feel like a package deal the CFO won’t unwind, steep multi-year discounts on renewals that happen to land in the middle of a challenger’s evaluation, and FUD from the consulting partners who have nine-figure Workday practices to protect. Contractual friction on data portability when customers try to export their tenant, and accelerated M&A where a challenger gets real traction. None of these moves address the underlying architectural problem, but any one of them can slow a design-partner deal by a quarter, and a challenger that hasn’t priced in the fight will burn runway on it. The thesis survives not because Workday won’t fight, but because the architecture they’re fighting with can’t be rebuilt into what F500 customers will actually need.
The opportunity
HR software is one of the last remaining enterprise categories where the incumbent is vulnerable, the architecture needs to be rewritten, and the buyer actively wants an alternative. The global HCM software market is more than $40 billion and growing, and Workday alone was worth close to $80 billion at its peak just two years ago. We think the AI transition creates an even bigger company.
And the stakes are bigger than the opportunity. As companies move to a workforce that’s a mix of humans and agents operating across the same systems, the HRIS becomes the substrate for how those companies are actually run — who reports to whom, who has what permissions, who gets paid, who’s accountable, who’s compliant. Building that on top of a 2005 architecture caps how much AI any company can actually deploy.
Right now, somewhere, an HR admin is copying seventeen comp adjustments from a spreadsheet into a Workday performance cycle, one field at a time, while the business partner watches on Zoom to make sure she doesn’t pick the wrong job code. That work is happening in every Fortune 500 today, on a product that costs the company multiple millions a year. Someone is going to build the next Workday — a system built for agents, not forms and approvals. Once that happens, no one is going back.
If you’re building this, reach out to jschmidt@a16z.com.
