Hebbia
B2B subscription SaaS, per-seat + platform pricing (Pro seats ~$10K/yr, Lite ~$3-3.5K/yr; avg contract value ~$500K), sold direct to financial services, consulting and legal enterprises
Only the Series B ~$700M mark is company-disclosed; Seed and Series A valuations were not published and are round-implied estimates. No Series C or secondary priced round had been disclosed as of the vintage (mid-2026), so the trail ends at the Jul-2024 Series B.
Earnings, margins, COGS & capex
Hebbia is a fast-scaling private vertical-AI SaaS. The last company-confirmed hard number is ~$13M ARR at the Jul-2024 Series B (up ~15x in 18 months), and management stated the business was profitable at that raise - unusual for an AI startup. Post-2024 revenue is only available as third-party estimates (~$30M ARR by end-2024 per Sacra; 2025 figures diverge widely and are unconfirmed), which should be treated as directional. Economics hinge on ~$500K average contract values in financial services and high net revenue retention (third-party figures range from ~140% to a circulated ~200%, neither company-audited). The main cost lever is LLM inference (OpenAI/Anthropic), so gross margin is a function of model-cost efficiency versus per-seat price.
Revenue trend
Margins
pressured by LLM inference COGS; improves as model costs fall
unknown - likely pushed back toward burn as it scales GTM and headcount
COGS structure
Dominated by third-party LLM API/inference spend (OpenAI, Anthropic) and cloud hosting (Microsoft Azure / hyperscaler compute), plus data-connector and storage costs. Not a hardware/capex cost base.
Capex
Minimal - asset-light software company; no owned data centers. Compute is rented from hyperscalers and inference bought from model providers.
Latest earnings
n/a
No public guidance. Management framing (2026 blog/press) centers on deterministic multi-agent 'agent swarms' and closing the individual-vs-firm productivity gap.
- ARR (Jun 2024, disclosed)
- ~$13M
- Series B valuation
- ~$700M (54x disclosed ARR)
- Avg contract value
- ~$500K
- Top asset managers by AUM as customers
- ~33-40%
- Pages processed
- 1B+ (company claim)
Growth drivers
- Deep penetration of financial services — claims ~33-40% of the largest global asset managers by AUM as customers, expanding seat count within each account
- Land-and-expand — high average contract value (~$500K) and strong net revenue retention within banks, PE/credit funds and consulting firms
- Vertical expansion beyond finance into legal (law firms like Fenwick, Seyfarth Shaw) and government (US Air Force)
- Matrix multi-agent platform + the Jun-2025 FlashDocs acquisition extending from analysis into automated deliverable generation (memos, decks)
- Secular shift to 'agent employees' automating repetitive finance/legal document workflows
Bull & bear
Hebbia is an early leader in agentic AI for the highest-value document work on Wall Street, with an elite, expanding customer base, ~$500K ACVs, strong retention, and a rare profitable-at-scale posture - a credible category winner in a market where comparable names (AlphaSense ~$7.5B, Glean ~$7.2B, Harvey ~$11B) now carry multi-billion marks well above Hebbia's stale $700M.
- Land-and-expand inside asset managers and banks with high ACV and high net retention compounds revenue fast off a small base
- Profitability at the Series B signals real unit economics, not just growth-at-all-costs - optionality to raise at a large step-up or stay lean
- Matrix + FlashDocs extends TAM from 'search/analysis' to 'produce the deliverable,' capturing more of the analyst workflow
- Vertical AI with proprietary retrieval and deep domain trust is defensible against generic horizontal copilots
- Private-market comps (AlphaSense $7.5B, Glean $7.2B, Rogo ~$2B, Harvey $11B) have all re-rated sharply, suggesting the two-year-old $700M mark is likely stale to the upside on a next round
A ~$700M mark on ~$13M disclosed ARR (54x) bet on years of flawless hypergrowth in a market being invaded simultaneously by foundation-model labs, terminal incumbents and a swarm of equally funded startups - with Hebbia's core reasoning rented from the very labs that could disintermediate it. And the loud re-rating of rivals cuts both ways: it may reflect the category winners pulling away from Hebbia, not just a stale mark.
- 54x entry multiple leaves no room for a growth stumble; even if ARR reached ~$30M+ it is still a fragile base for a mark that must keep compounding
- Dependence on OpenAI/Anthropic caps margin control and exposes Hebbia to being out-shipped by the model providers' own enterprise agents
- Direct competitors (Rogo ~$2B, AlphaSense ~$7.5B, Glean ~$7.2B, Harvey ~$11B) are far better-capitalized and overlapping; differentiation may erode to a feature race Hebbia is now outspent in
- Incumbents (Bloomberg, S&P, FactSet, Microsoft) own the distribution and data Hebbia must sit alongside and can bundle 'good-enough' AI
- Revenue transparency is limited post-2024 - the growth story rests on company claims and diverging third-party estimates, not audited numbers
What it is worth
Private-market: last priced round (Series B, ~$700M at ~$13M ARR = 54x, Jul 2024) triangulated against private AI-SaaS comps (AlphaSense ~$7.5B, Glean ~$7.2B, Harvey ~$11B, Rogo ~$2B, all 2025-2026 marks) and forward ARR - no public price or DCF.
~$400-700M or a flat/down round
if growth decelerated, foundation labs encroach, better-capitalized rivals pull ahead, and the 54x entry multiple compresses; small disclosed revenue base leaves little cushion.
~$1-1.5B
last mark re-rated for 2025 growth and a broadly re-priced vertical-AI comp set; premium vertical AI SaaS multiple on an estimated ~$30M ARR base.
~$2-3B+ on a Series C
if ARR is compounding toward/through ~$50M+ with retained >150% net retention and comps hold - in a range now bracketed by AlphaSense/Glean/Harvey-class marks.
The $700M mark is ~2 years stale and predates most 2025 growth. On the ~$13M disclosed ARR it was 54x; if third-party ~$30M+ ARR estimates are near-right, the same mark is well under 30x forward, and comparable private AI names have re-rated to multi-billion valuations - implying a likely step-up on a next round. That re-rating cuts both ways: it may also signal the category winners pulling ahead of Hebbia. The multiple demands durable hypergrowth and margin defense against foundation-model encroachment; treat all post-2024 figures as estimates.
SWOT
Strengths
- Elite customer base — banks, PE/credit, top consulting firms and a claimed ~33-40% of the largest asset managers by AUM; sticky, high-ACV enterprise relationships
- Tier-1 backing (a16z, Index, Google Ventures, Peter Thiel) and a strong founder narrative (George Sivulka, ex-Stanford PhD)
- Was profitable at the Series B raise - capital discipline rare among AI startups
- Proprietary iterative source-decomposition / multi-step retrieval that avoids naive chunking, tuned for large proprietary document sets and data rooms
Weaknesses
- Small absolute revenue (~$13M disclosed ARR) against a ~$700M mark - a 54x multiple that needs continued hypergrowth to justify
- Heavy dependence on third-party foundation models (OpenAI/Anthropic) for the core reasoning layer - margin and capability are partly outside its control
- Concentrated in financial services; vertical/geographic diversification still early
- Crowded, well-funded competitive set (AlphaSense, Rogo, Glean, Harvey) plus incumbents (Microsoft, Bloomberg, S&P/FactSet) that can bundle
Opportunities
- Expand from analysis into full workflow automation (deliverables, models, memos) via FlashDocs and Matrix agents
- Verticalize into legal, consulting, government and other document-intensive industries
- Ride falling inference costs to expand gross margin while holding seat price
- Displace legacy research/terminal spend (Bloomberg, FactSet, Capital IQ) as agentic research matures
Threats
- Foundation-model providers (OpenAI, Anthropic, Microsoft Copilot) moving up-stack into enterprise agents and document reasoning - disintermediation risk
- Incumbent data/terminal players (Bloomberg, S&P Global, FactSet, AlphaSense) embedding comparable AI into existing distribution
- Model-cost or model-access shocks compressing margins
- Enterprise data-security/compliance scrutiny in finance and legal slowing adoption or raising cost-to-serve
Moats, dependencies & bottlenecks
Moats
Deep integration into due-diligence and research workflows at elite firms creates switching costs; but trust is contestable by incumbents with existing relationships.
Proprietary retrieval / multi-step decomposition over large proprietary corpora Real technical edge for un-chunked, iterative analysis, but the reasoning layer is a third-party LLM - the defensible IP is orchestration, not the model.
1B+ pages processed and account-specific agent tuning improve product, but customer data is siloed and not pooled, limiting network effects.
a16z/Thiel backing and marquee logos aid GTM, but signaling is not a durable economic moat.
Dependencies
Core technology / COGS The reasoning engine is rented; pricing, access and capability shifts hit both margin and product, and these providers are potential competitors.
Standard hosting dependency; commoditized but a real cost line.
Heavy reliance on asset managers/banks; a sector budget pullback or a few large-account losses would materially dent ARR.
Profitable at last raise, but category land-grab likely requires more capital; a hard funding market would constrain GTM.
Advantages
- Early mover with elite, high-ACV financial-services customer base and strong claimed retention
- Profitable-at-scale posture giving capital optionality
- Full-workflow ambition (analysis -> deliverable) via Matrix + FlashDocs
- Founder/brand and top-tier investor and logo credibility for enterprise GTM
Weaknesses
- High-multiple valuation dependent on sustained hypergrowth
- Reasoning layer outsourced to potential competitors
- Limited post-2024 financial transparency
- Narrow vertical concentration and a crowded, better-capitalized competitive field
Bottlenecks
- Gross margin governed by third-party inference costs Hebbia doesn't control
- Enterprise sales cycles and security/compliance reviews in finance and legal are long and expensive
- Talent competition for applied-AI and vertical GTM against far-larger, better-funded labs and incumbents
- Data-access and integration friction inside heavily regulated customer environments
Top signals & trends
Top signals
Given AlphaSense (~$7.5B), Glean (~$7.2B) and Harvey (~$11B) comps and Hebbia's growth, a large up-round would validate the trajectory; absence of one would raise questions.
A confirmed figure well above the ~$13M print would de-risk the multiple; only third-party estimates exist today.
OpenAI/Anthropic/Microsoft moving up-stack is the key disintermediation catalyst to watch.
Fenwick, Seyfarth Shaw, US Air Force wins show TAM beyond finance.
Signals platform expansion from search into deliverable/slide generation.
Trends
Hebbia's 2026 positioning is squarely on institutional, checkpoint-based agents automating whole workflows.
Directly expands gross margin if seat pricing holds.
Compresses differentiation and raises disintermediation risk.
Expands the spend pool Hebbia can capture from analyst labor and legacy research tools.
Raises cost-to-serve but favors trusted, purpose-built vendors over generic tools.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Core LLM reasoning provider; also a case-study partner (Hebbia featured by OpenAI).
Foundation-model provider for reasoning workloads.
Cloud hosting / compute for the platform.
Underlying GPU compute behind the model/inference layer (indirect).
Vector-database category tooling in the retrieval stack (ecosystem-level).
Investment bank; named due-diligence/research customer.
Credit/alternative asset manager customer.
Private equity customer.
Private equity customer.
Law firm - legal-vertical customer.
Law firm; processed 7M+ pages (company claim).
Government customer - vertical expansion signal.
Claims ~33-40% of the largest global asset managers by AUM as customers; specific names largely undisclosed.
Market-intelligence + AI search over premium content library; raised $350M at ~$7.5B valuation (Jun 2026), $600M+ ARR. Overlaps on financial research but content-library-centric vs Hebbia's proprietary-document focus.
AI agents generating IB deliverables (memos, comps, pitch decks); Series C at $750M (Jan 2026), reportedly ~$2B on a subsequent round. Fast-growing direct competitor for banking/finance workflows.
Horizontal enterprise AI search/assistant; $150M Series F at ~$7.2B valuation (Jun 2025), $100M+ ARR. Broader but less finance-specialized than Hebbia.
Vertical AI for legal; ~$11B valuation (Mar 2026). Overlaps as Hebbia pushes into legal document work.
Bundled enterprise AI + distribution + Azure; 'good-enough' copilots as a bundling threat.
Terminal incumbent embedding AI; owns finance distribution and data Hebbia sits alongside.
Structured financial data + comps with growing AI features; incumbent research spend Hebbia targets.
Multi-asset analytics incumbent adding AI; competes for the research/analyst budget.
Both critical suppliers and potential up-stack competitors if they ship native enterprise document agents.