Cresta
Enterprise B2B SaaS — per-agent-seat subscriptions plus feature tiers (real-time assist, conversation intelligence, QM, autonomous virtual agents); annual contracts, land-and-expand via seat growth and AI-automation upsell.
Series C (Mar 2022, $1.6B) is the last officially priced round. The Nov 2024 Series D did not disclose a valuation; third-party trackers carry the mark roughly flat at ~$1.6B, so the 2024 valuationB is a tracker estimate, not a disclosed priced mark. Earlier rounds (Series A 2020, Series B $50M Mar 2021) had no publicly disclosed valuation and are omitted to avoid fabricated marks.
Earnings, margins, COGS & capex
Cresta is a growth-stage private company with no public financials. It has raised over $270M across its disclosed rounds (Series B $50M Mar 2021, Series C $80M Mar 2022 at $1.6B, Series D $125M Nov 2024). The only company-disclosed operating metrics are directional: ARR 'nearly quadrupled' and customers 'nearly doubled' in the ~2 years to the Series D. The primary third-party ARR estimate is ~$52M (GetLatka); treat all point revenue figures as unverified estimates.
Revenue trend
Margins
structurally pressured by per-interaction LLM inference cost
presumed negative, growth-stage
n/a
COGS structure
Not disclosed. Cost of revenue is dominated by real-time inference/model compute (Cresta runs AI on every live call/chat/email) plus cloud hosting and customer-success delivery for large enterprises. Real-time LLM usage on high-volume contact-center traffic is a distinctive, usage-scaling COGS line that separates Cresta's unit economics from classic seat-based SaaS.
Capex
Not disclosed. Asset-light — no owned data centers implied; compute is rented cloud/model capacity. Investment is R&D and GTM headcount, with new engineering hubs announced in Romania and India plus offices in Palo Alto, San Francisco, New York, Berlin, Toronto.
Latest earnings
n/a
n/a — no public guidance
- Total funding raised
- >$270M (through Series D)
- Last round
- $125M Series D, Nov 19 2024
- Last disclosed valuation
- $1.6B (Series C, Mar 2022); Series D valuation not disclosed
- Named enterprise customers
- Intuit, Verizon, CarMax, Cox Communications, Hilton, Brinks Home
Growth drivers
- Shift from real-time human-agent assist toward autonomous/virtual AI agents (higher-value, automation-priced expansion beyond seat count)
- Enterprise generative-AI adoption in customer service — large logos consolidating point tools onto an AI copilot layer
- Land-and-expand — seat growth plus feature upsell (conversation intelligence, QM, agent automation) within existing accounts
- Distribution leverage from strategic investors/partners (Accenture as SI channel; embeds with Genesys, Five9, Amazon Connect, Salesforce, Zendesk, ServiceNow)
- International expansion and lower-cost engineering hubs (Romania, India) to scale delivery
Bull & bear
Cresta is a full-stack, generative-AI-native contact-center platform with real enterprise traction and a strategic cap table, riding the shift from human-agent assist to autonomous virtual agents — the largest near-term automation budget in the enterprise.
- Company-stated ARR nearly quadrupled in ~2 years with a nearly-doubled customer base — genuine enterprise pull, not a demo-stage company
- Suite breadth (assist + intelligence + QM + virtual agents) positions it to capture both the seat-based assist budget AND the far larger call-deflection/automation budget
- Strategic investors double as channel: Accenture (SI), Qualcomm, Workday, LG, plus platform embeds (Genesys, Five9, Amazon Connect, Salesforce, ServiceNow)
- Marquee, sticky logos (Intuit, Verizon, Hilton, CarMax, Cox, Brinks Home) with high switching costs once embedded in agent workflows
- $270M+ raised gives multi-year runway to outlast weaker point-solution rivals through the AI-spend shakeout
Cresta is a best-of-breed overlay competing against both well-capitalized CCaaS incumbents bundling AI for free and hyperscalers with distribution — with a roughly-flat $1.6B valuation, undisclosed (probably deep) losses, and inference-heavy gross margins.
- Valuation appears roughly flat since 2022 ($1.6B) despite a large 2024 raise — a warning sign on pricing power and investor markups
- The core risk is bundling: NICE, Five9, Genesys, Talkdesk and Amazon Connect can ship 'good-enough' AI at near-zero incremental price into installed bases Cresta must win one deal at a time
- Sits on other people's LLMs and other people's CCaaS platforms — thin structural control; model commoditization erodes differentiation
- Real-time inference on every interaction structurally caps gross margin below pure SaaS and worsens with volume
- Very crowded field (Observe.AI, ASAPP, Level AI, Balto, Cogito) plus undisclosed burn — continued dependence on capital markets that may be less friendly at a flat mark
What it is worth
Private-market last-round anchor + ARR-multiple sanity check (no public comps applied directly; peers NICE/FIVN/VRNT used only for multiple context).
<$1B in a down-round or distressed-M&A outcome if bundling compresses growth/margins and capital markets stay tight — a plausible flat-to-2022 mark masking a real per-share decline.
~$1.6B
roughly flat with the last disclosed mark — consistent with a $125M raise that did not publicly step the valuation up in a reset software market.
~$2.5-3.5B+ in an IPO/strategic-exit scenario if ARR materially exceeds the ~$52M estimate and reaccelerates on virtual-agent automation, warranting a premium AI-native growth multiple.
Last disclosed priced valuation is $1.6B (Series C, Mar 2022, which 'quadrupled' the prior mark); the $125M Series D (Nov 2024) did not publicly disclose a new valuation and third-party trackers carry the mark roughly flat at ~$1.6B. On GetLatka's ~$52M ARR estimate, that implies a rich ~30x ARR multiple — well above listed CX-software peers (NICE/FIVN/VRNT trade at low-to-mid single-digit revenue multiples). All ARR figures are unverified estimates; the true multiple is unknowable without disclosed financials.
SWOT
Strengths
- Full-stack contact-center AI suite (real-time assist + conversation intelligence + QM + virtual agents) rather than a single point feature
- Blue-chip enterprise logos (Intuit, Verizon, CarMax, Hilton, Cox, Brinks Home) validating deployment at scale
- Deep-pocketed, strategic cap table — a16z, Greylock, Sequoia, Tiger, J.P. Morgan, plus Accenture, Qualcomm, Workday, QIA, WiL, LG — providing >$270M runway and distribution
- Founder/technical pedigree (Sebastian Thrun, Zayd Enam, Tim Shi) and an early real-time on-call ML focus
Weaknesses
- No disclosed path to profitability — heavy cash burn typical of growth-stage AI, dependent on continued fundraising
- Real-time inference on every interaction is a structural gross-margin drag vs classic SaaS
- Valuation appears roughly flat since 2022 ($1.6B) — a possible flat-round signal amid the 2022-24 reset in software multiples
- Application layer sitting on third-party LLMs and third-party CCaaS platforms it does not control
Opportunities
- Autonomous virtual-agent automation monetized on outcomes/volume, decoupling revenue from human seat counts (larger TAM than assist)
- Displacing legacy QM/analytics incumbents (Verint, NICE) with generative-AI-native tooling
- SI-led enterprise expansion via Accenture and platform marketplaces (Amazon Connect, Genesys, Salesforce)
- International and vertical expansion (regulated verticals — financial services, insurance, telecom, hospitality)
Threats
- Platform consolidation — CCaaS incumbents (NICE, Five9, Genesys, Talkdesk) and hyperscalers (Amazon Connect, Google, Microsoft) bundling AI at near-zero marginal price
- LLM commoditization eroding the model-level moat as base capabilities become table-stakes
- Crowded overlay field (Observe.AI, ASAPP, Level AI, Balto, Cogito) compressing pricing
- Enterprise preference for single-vendor stacks squeezing best-of-breed point solutions
- Macro/ROI scrutiny on AI spend elongating enterprise sales cycles
Moats, dependencies & bottlenecks
Moats
Once Cresta is wired into agent desktops, QM, and real-time coaching across thousands of seats, ripping it out is costly — but the workflow can be re-hosted by an incumbent platform over time.
Volume of real-time enterprise conversation data enables domain-tuned models; partially eroded as base LLMs improve and rivals accumulate comparable data.
End-to-end assist+intelligence+QM+virtual-agent reduces the need for multiple vendors, but competes head-on with incumbents that own the underlying CCaaS platform.
SI and marketplace channels accelerate enterprise reach; not exclusive and replicable by rivals.
Thrun/Stanford-AI halo and marquee logos aid credibility, but do not by themselves defend pricing.
Dependencies
Anthropic, Google, AWS Bedrock) Technology / supply Core AI capability and a major COGS line depend on third-party LLMs; pricing, availability, and capability shifts flow straight through.
Hosting and inference compute; concentration and cost exposure — and hyperscalers are also competitors via Amazon Connect / Google.
Integration / channel Cresta overlays these telephony/routing platforms; each is simultaneously an integration partner and a potential bundling competitor.
Large-contract sales cycles are sensitive to macro and AI-spend scrutiny.
Undisclosed burn implies continued reliance on equity funding; a flat mark raises the cost/difficulty of the next round.
Advantages
- End-to-end suite spanning human-agent assist and autonomous virtual agents in one platform
- Real-time (on-call) intelligence heritage rather than post-call analytics only
- Enterprise-grade reference customers across telecom, retail, hospitality, fintech, and home security
- Strategic, well-capitalized cap table providing runway and distribution leverage
Weaknesses
- No disclosed profitability path; presumed significant cash burn
- Inference-heavy cost structure caps margins vs classic SaaS
- Application-layer position atop third-party models and CCaaS platforms
- Valuation roughly flat since 2022 despite continued raising — weak markup signal
- Crowded competitive field pressuring differentiation and price
Bottlenecks
- Gross-margin ceiling from real-time per-interaction inference cost that scales with usage
- Long, complex enterprise sales cycles and heavy customer-success delivery for large deployments
- Dependence on third-party LLMs and CCaaS platforms it does not control
- Talent and compute cost inflation in a competitive AI hiring/inference market
- Proving durable ROI/automation outcomes to justify premium pricing against free-bundled incumbent AI
Top signals & trends
Top signals
Fresh capital and strategic backers validate the platform and extend runway.
An undisclosed / flat mark across a large 2024 raise suggests limited pricing power / cautious investors.
Directional but real growth; not independently verifiable and no absolute ARR disclosed.
Cost discipline / margin management consistent with a path toward efficiency.
Structural pricing threat to the overlay category.
Trends
Expands TAM from seat-based assist to call-deflection/automation budgets — Cresta's key growth vector and battleground.
Tailwind for the whole category; also invites incumbent and hyperscaler entry.
Primary threat — incumbents package AI at near-zero marginal price into installed bases.
Lowers barriers to entry and erodes model-level differentiation; shifts moat to data, workflow, and outcomes.
Buyers demand measurable deflection/efficiency, lengthening cycles but rewarding proven outcomes.
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.
Foundation-model provider powering generative capabilities.
Foundation-model provider (Claude) for enterprise-grade LLM capability.
Model + cloud supplier; also a competitor via Google CCAI.
Cloud infrastructure and model access; also a competitor via Amazon Connect.
Underlying GPU/inference compute layer behind hosted models.
Named enterprise customer (fintech / SMB software).
Named enterprise customer (telecom) — cited in the Series D release.
Named enterprise customer (auto retail).
Named enterprise customer (hospitality).
Named enterprise customer (telecom / cable).
Named enterprise customer (home security).
CCaaS/CX incumbent (CXone) with native generative-AI (Enlighten) and QM — the archetypal bundling threat; also owns conversational-AI assets. Dual-listed on NASDAQ (NICE) and Tel Aviv.
Cloud contact-center platform embedding AI agents and agent assist directly into its stack — both partner and competitor.
Workforce-engagement and QM incumbent pushing generative-AI bots; competes on conversation intelligence and QM.
Owns the CRM system of record and is pushing autonomous service agents (Agentforce) — a large-platform bundling threat and an integration partner.
Hyperscaler CCaaS with native AI (Q in Connect); distribution + cost advantage; Cresta both integrates with and competes against it.
Dynamics 365 Contact Center + Copilot bringing AI service agents to enterprises on the Microsoft stack.
Customer-engagement/communications platform adding AI agent capabilities; adjacent competitive surface.
Major CCaaS incumbent (integration partner) embedding native AI; strong installed base makes bundling a direct threat.
Cloud contact-center platform with its own AI agents; overlaps in AI assist/automation.
Direct overlay rival — conversation intelligence, agent assist, and QM for contact centers.
Generative-AI contact-center automation serving Fortune-100 CX; direct best-of-breed competitor.
AI-native QM/conversation-intelligence startup competing for the same enterprise budget.
Real-time agent-guidance specialist — direct overlap with Cresta's real-time assist.
Real-time coaching / emotion-AI for contact centers — overlapping assist category.