Should ServiceNow acquire UiPath to win agent automation?
No. ServiceNow should not acquire UiPath at any price north of $3B, and probably should not acquire it at all. UiPath is a RPA-heritage company in a forced pivot to AI agents, with its market cap collapsed from $30B+ peak (2021) to $5-7B (Q1 2026), revenue growth decelerated to 8-10%, and named-customer churn signals leaking through earnings calls.
The category UiPath defined — desktop/screen-scraping RPA — is being eaten alive by Microsoft Power Automate Premium ($15/user/mo, free-with-M365 thesis) and Copilot Actions, which means ServiceNow would be paying $5-7B + 18-month integration drag for a melting ice cube. The only scenario where it makes sense: UiPath drops below $3B in a 2027 distress event, ServiceNow does an Activision-style structured deal, and the real prize becomes the 800-engineer EU automation team plus the Fortune 500 install base — not the product.
Until then, ServiceNow's AI Agent Studio + organic build + $300-500M tuck-ins (Decagon-tier) is the dominant strategy.
The UiPath Reality In 2026
- Market cap collapse: From $30B+ peak (May 2021) to $5-7B as of Q1 2026 — an 80%+ drawdown that reflects investor verdict on RPA-as-a-category, not just execution
- Revenue deceleration: ~$1.4B FY25 ARR growing 8-10% YoY, down from 40%+ growth in 2021-22; net retention has fallen from 144% (2021) to ~110% (FY25)
- AI agent pivot status: UiPath Autopilot + Agent Builder launched 2024-25 with mixed reception — analyst notes flag "product-market fit unclear" and overlap with native Microsoft/ServiceNow agent stacks
- Named churn signals: Multiple Fortune 500 deployments downsizing seat counts as Power Automate Premium absorbs use cases; Gartner MQ position weakening vs. Microsoft
- CEO/leadership churn: Daniel Dines returned as CEO in 2024 after Rob Enslin departure — classic founder-rescue pattern that often precedes either turnaround or distress sale
The 4 Reasons NOT To Buy UiPath
- The RPA category is being eaten by Microsoft: Power Automate Premium at $15/user/mo (and increasingly bundled with M365 E5 + Copilot) is destroying standalone RPA pricing. ServiceNow would be acquiring a category in structural decline, not a growth asset.
- It dilutes ServiceNow's AI-first narrative: ServiceNow has spent 24 months positioning AI Agent Studio + Now Assist as the workflow-native AI control plane. Bolting on a screen-scraping RPA vendor is a step backward in narrative — Wall Street would mark it as defensive M&A.
- $5-7B is wildly overpriced for a transitioning company: At ~4-5x revenue for a sub-10% grower with declining net retention, this is a premium SaaS multiple for a company that should trade at infrastructure-software multiples (2-3x). The fair price is closer to $2.5-3.5B.
- Cultural and architectural mismatch: UiPath is a Windows-desktop, agent-on-endpoint vendor. ServiceNow is a multi-tenant cloud workflow platform. Integrating UiPath's runtime into the Now Platform is a 24-36 month engineering slog with high risk of stranded investment.
The 1 Scenario Where It Could Work
- Distress trigger: UiPath market cap drops below $3B in a 2027 down round / activist pressure event (Elliott or Starboard already circling per industry chatter)
- Structured deal mechanics: ServiceNow does a Microsoft-Activision-style structured acquisition — cash + earnout + retention pool, total enterprise value $2.5-3.2B, with regulatory carve-outs for the EU agent runtime
- Acquihire thesis: The 800-person EU automation engineering team (Bucharest, Bellevue) becomes the real asset — folded into ServiceNow's AI Agent Studio org under a single VP
- Customer-base play: UiPath's Fortune 500 install base (Coca-Cola, GE, Cisco) gets migrated onto Now Platform within 18 months — net new ServiceNow workflow seats become the ROI justification
- Product sunset path: UiPath standalone product gets a 36-month maintenance commitment, then end-of-life as customers migrate to native AI Agent Studio
What ServiceNow Should Do Instead
- Build out AI Agent Studio organically: Already 18 months ahead on workflow-native agents; double engineering headcount to 600+ on the agent runtime, ship monthly
- Acquire smaller AI-native ($300-500M Decagon-tier): Decagon, Sierra, Ada, or similar AI-customer-service agents — workflow-native, no RPA legacy debt, integrate in 90 days vs. 24 months
- Partner with named RPA legacy customers: Direct outbound to UiPath's top-50 accounts with a "Now Platform AI Agent Studio + 50% RPA seat-cost" replacement offer
- Talent acquisition from UiPath alumni: Hire 50-100 senior engineers from UiPath via individual offers — same engineering capability, 2% of the cost, no integration drag
- OEM Microsoft Power Automate (counterintuitive): If you can't beat them, embed Power Automate triggers natively in Now Platform — turns Microsoft into a feeder, not a competitor
- Quiet partnership with UiPath itself: Buy the integration, not the company — joint go-to-market, revenue share, optionality on future acquisition without the balance-sheet risk
The Microsoft Question
- Power Automate Premium pricing kills standalone RPA: $15/user/mo is 60-70% below UiPath equivalent; bundled with M365 E5 it's effectively free for the largest enterprise buyers
- Copilot Actions absorbs the agent layer: Microsoft is shipping native AI agents in Teams, Outlook, and Dynamics — the "automation" use case is being absorbed into the productivity suite
- The free-with-M365 thesis: Once a category is bundled into a $57/user/mo productivity suite, standalone vendors lose pricing power permanently — see what happened to standalone email security vs. Defender
- Named pattern: Same dynamic killed standalone web conferencing (vs. Teams), standalone diagramming (vs. Visio), standalone forms (vs. Forms) — RPA is next
- Implication for ServiceNow: Acquiring UiPath means inheriting a price war you cannot win against an opponent who can give the product away
The Comparable Failed RPA-Adjacent Acquisitions
- BMC + Innotas (2018): PPM tuck-in that took 36 months to integrate, shed 40% of customers, written down as goodwill in 2021
- Pega + various RPA tuck-ins (2019-22): Pega's RPA strategy — partly organic, partly acquisition — never produced material revenue lift; analyst day disclosures dropped RPA as a separate category in 2024
- Appian + Jitterbit partnership (not acquisition, but instructive): Appian wisely chose partnership over acquisition for integration capability — preserved balance sheet, kept narrative clean
- Salesforce + MuleSoft (2018): $6.5B price, integration took 4 years, eventually justified — but only because integration-platform-as-a-service grew into AI; RPA has no such tailwind
- Pattern: Workflow platform + RPA vendor acquisitions consistently underperform deal models by 30-50%; the integration drag and narrative confusion outweigh revenue synergy
Strategy Comparison Table
| Strategy Option | Cost | Strategic Fit | Risk | Recommendation |
|---|---|---|---|---|
| Acquire UiPath at current $5-7B | $5-7B + integration | Low (RPA legacy) | Very High | NO |
| Acquire UiPath in 2027 distress at <$3B | $2.5-3.2B structured | Medium (acquihire) | High | Maybe |
| Acquire AI-native ($300-500M Decagon-tier) | $300-500M | High (workflow-native) | Low | YES |
| Build AI Agent Studio organically | $200M/yr R&D | Very High | Low | YES (already doing) |
| OEM Microsoft Power Automate | Revenue share | Medium (frenemy) | Medium | Consider |
| Partner with UiPath, no acquisition | Minimal | Medium | Low | YES (interim) |
| Talent raid UiPath alumni (50-100 engineers) | $30-50M/yr | High | Low | YES |
Strategic Decision Flow
FAQ
How far has UiPath's market cap fallen? From a $30B+ peak in May 2021 to roughly $5-7B as of Q1 2026 — an 80%+ drawdown that reflects the investor verdict on RPA as a category, not just execution. Revenue growth decelerated to 8-10% YoY from 40%+ in 2021-22, and net retention fell from 144% in 2021 to about 110% in FY25.
Why is Microsoft the core reason not to buy? Power Automate Premium at $15/user/mo is 60-70% below UiPath equivalent, and bundled with M365 E5 it is effectively free for the largest enterprise buyers. Once a category is bundled into a productivity suite, standalone vendors lose pricing power permanently — the same dynamic that killed standalone web conferencing versus Teams and standalone diagramming versus Visio.
What price would actually be fair for UiPath? The article argues $5-7B is wildly overpriced for a sub-10% grower with declining net retention. At a fair infrastructure-software multiple of 2-3x revenue, the price should be closer to $2.5-3.5B, and ServiceNow should not pay north of $3B.
What is the one scenario where acquiring UiPath works? A 2027 distress trigger drops UiPath below $3B amid activist pressure (Elliott or Starboard chatter), and ServiceNow does a Microsoft-Activision-style structured deal at $2.5-3.2B enterprise value. The real prize becomes the 800-person EU automation engineering team in Bucharest and Bellevue, plus migrating the Fortune 500 install base (Coca-Cola, GE, Cisco) onto the Now Platform within 18 months.
What should ServiceNow do instead of buying UiPath? Build AI Agent Studio organically by doubling agent-runtime engineering to 600+, acquire smaller AI-native customer-service agents in the $300-500M Decagon tier that integrate in 90 days, and hire 50-100 senior engineers directly from UiPath at about 2% of acquisition cost with no integration drag.
A quiet partnership to buy the integration rather than the company preserves optionality without balance-sheet risk.
Bottom Line
No — ServiceNow should not acquire UiPath at $5-7B. The RPA category is structurally declining under Microsoft Power Automate pressure, the integration would dilute ServiceNow's AI-first narrative, and the price is 1.5-2x what a transitioning company should command. The dominant strategy is organic AI Agent Studio investment + $300-500M AI-native tuck-ins + talent raid on UiPath alumni.
The only scenario where acquisition makes sense is a 2027 distress event at <$3B with a structured Activision-style deal where the engineering team and customer base — not the product — are the prize.
*(see also: q1620, q1628, q1655)*