Ema Just Raised $77M to Kill Enterprise SaaS — ServiceNow, Salesforce and Workday Posted Record AI Quarters Anyway. Who's Actually Winning?

Ema, a 2-year-old San Francisco startup, announced a $77 million Series B on September 23, 2026, led by Bengaluru-based Creaegis, with existing backers Accel, S32 and Prosus all increasing their stakes. The round brings Ema's total funding to $140 million and more than quadruples its valuation from the 2024 Series A, according to the company's announcement.

The pitch is unusually blunt for a B2B round. Ema sells 'AI Employees' — multi-agent systems that complete entire workflows in HR, IT and finance rather than help humans click through SaaS screens. Pricing is tied to outcomes, not seats or tokens. The company says it has more than 50 active enterprise deals, 1 million+ users, 5 million+ agent actions and queries handled, and bookings of $150 million+ across multi-year contracts.

Within the same 90-day window, the public companies Ema's thesis most directly threatens printed their strongest AI results ever. The numbers, all from the most recent 10-Q or earnings release for each company, make the displacement case harder than the round's framing suggests.

All three incumbents added AI products faster than their headline growth rate. Salesforce's Agentforce alone — launched in 2024 — hit $1.5B ARR within roughly18 months, a pace that took legacy Salesforce products nearly a decade. ServiceNow reports AI control-tower deployments up ninefold in nine months and more than 500 live customers. Both stocks have been among the market's best AI trades of the past quarter: Salesforce jumped 22.6% on its earnings day; ServiceNow is up more than 25% over the trailing month.

The data points to a more nuanced outcome than either side of this debate expects. Ema's traction is real — 50x revenue growth, $150M+ in bookings, ~80% gross margins, ~180% NDR — but those numbers are off a tiny base in markets where the incumbent has $10B+ of installed contracts and the workflows already run on its database. The $77M is a venture bet on a future state, not a current revenue line that threatens ServiceNow's $3.99B quarter.

The deeper pattern is that agentic AI is splitting the public SaaS stack into three tiers, with the round's $77M acting as a marker for the dividing lines rather than a knockout blow.

The supply-chain lens sharpens the picture. Upstream of every Ema agent and every ServiceNow Now Assist workflow sit the same model providers — OpenAI (which launched its own Presence managed-agent product in July 2026) and Anthropic (private, which reported 300,000+ business customers in 2026) — and the same compute substrate from NVIDIA. Downstream, the customer logos are remarkably overlapping: Ema lists Microsoft, Alphabet, KPMG, PwC, Wipro, ADP, Hitachi and NTT DATA — many of the same Fortune 500 buyers that just renewed ServiceNow and Salesforce contracts.

Three observations on the stack from the chart above. First, NVIDIA's Q2 FY2027 revenue of $96.2B (ended July 26, 2026) is more than 8x Salesforce's, meaning every incremental agent action in any layer ultimately clears through inference compute. Second, Microsoft sits on both sides of the disruption: it is both a named Ema customer and the operator of Azure OpenAI, the model layer feeding most enterprise agents. Third, the services tier — Accenture at $18.7B Q3 FY26 revenue and advanced-AI bookings up 76% YoY — captures deployment labor for every agent in the stack, including Wipro's Ema deployment.

Three structural questions frame the1- to 3-year outlook. Gartner's August 2025 forecast — that 40% of enterprise applications will include task-specific AI agents by end of 2026, up from less than 5% in 2025 — is being met. The follow-on questions matter more.

Ema's $77M is the cleanest articulation yet of the agentic-AI thesis against the public SaaS stack — and the cleanest evidence that the thesis is incomplete. The market is not choosing between Ema or ServiceNow. It is choosing among three baskets, each priced for a different future.