HeyGen

HeyGen is a Los Angeles generative-video company founded in Shenzhen in December 2020 by former Snap engineer Joshua Xu and former Smule/ByteDance designer Wayne Liang. The product and corporate identity moved through Surreal, Movio and, from April 2023, HeyGen; the company relocated its headquarters to Los Angeles in 2022 and later dissolved its mainland operation. Its browser and API products turn scripts, images, audio or source video into presenter-led videos using stock avatars or consent-verified custom digital twins, voice clones, lip-synced translation and real-time interactive avatars. The platform serves a self-serve creator/SMB base and enterprise localization, marketing, enablement, learning and communications teams. Current plans run from a free tier through $29 Creator, $49-and-up Pro and $149 Business plans, with enterprise and separately metered API pricing. HeyGen says it reached profitability by Q2 2023 and grew ARR from $1m to more than $35m in roughly a year; by June 2026 CEO Joshua Xu reported $200m ARR, more than 30m users, 120m generated videos, 250+ languages, 85% of the Fortune 100, cash-flow break-even operations and only $25m of cumulative cash burn. These latest figures are first-party interview claims, not audited accounts, but imply exceptional capital efficiency: the company raised $74.65m, including roughly $9m of 2022 angel/Pre-A tranches from Chinese investors, a $5.65m 2023 Conviction seed and a $60m 2024 Benchmark-led Series A. The last round carried a roughly $477.4m precise preferred post-money estimate and was publicly described as valuing HeyGen above $500m. BOND, Conviction, Thrive, SV Angel and angels including Dylan Field and Elad Gil participated. The cap table was deliberately geopolitically restructured: early investors IDG Capital, Baidu Ventures, HongShan/Sequoia China and ZhenFund were asked to sell substantial stakes to U.S. investors, and Sarah Guo replaced HongShan on the board. Core differentiation is an integrated workflow spanning avatar creation, script-to-video, localization and API delivery with unusually strong lip sync and rapid self-serve adoption; major competitive sets include Synthesia, Colossyan, D-ID, Captions, Runway, OpenAI/Sora, Google/Veo and editing/localization suites. Material risks are synthetic-media abuse, consent and biometric-data compliance, copyright/likeness disputes, fast model commoditization, platform dependence and moderation at high volume. The Washington Post documented scammers using tools including HeyGen and ElevenLabs to manipulate stolen social-media footage for deceptive ads. HeyGen now requires actor consent for custom avatars, processes facial geometry to verify that consent, applies automated and human moderation, offers takedown/appeal paths, and reports SOC 2 Type II and GDPR controls; those controls reduce but cannot eliminate downstream misuse. A May 2026 account-takeover case study also alleged that password/MFA changes did not terminate existing sessions and that email changes lacked verification, a specific security-control concern rather than evidence of a platform-wide data breach. No material enforcement action or HeyGen-specific BIPA class action was found in this review.

Founding story

Xu and Liang studied at Tongji University and later Carnegie Mellon before working on U.S. consumer-video products. Xu spent about six years at Snap on ads ranking, machine learning and computational photography; Liang designed products at Smule and ByteDance. They founded Surreal in Shenzhen in December 2020 to remove the camera, studio and production bottleneck from professional video, then moved the company and product to Los Angeles as Movio and ultimately HeyGen.

Business model

Product-led SaaS and usage-based generative-video platform: free acquisition funnel, individual and team subscriptions, higher-touch enterprise contracts, and a separately funded API for programmatic generation and real-time avatars.

Monthly/annual seat and credit subscriptions plus enterprise contracts and pay-as-you-go API credits. Higher-resolution generation, premium models, custom twins, translation and team/enterprise controls consume plan credits or contracted capacity.

Traction

Official 2024 Series A materials reported more than 40,000 paying businesses, $35m+ ARR and profitability since Q2 2023. In June 2026 Joshua Xu reported $200m ARR, 30m+ users, 120m videos generated, 250+ supported languages and adoption by 85% of the Fortune 100; he said ARR doubled from $100m in eight months and the business was cash-flow break-even for 2026 to date.

Latest developments

As of June 2026 the company reported $200m ARR and cash-flow break-even operations, launched higher-end Avatar V/Video Agent workflows and continued expanding enterprise and developer offerings. Current plan architecture is credit-based, with Creator, Pro, Business and Enterprise tiers plus a standalone API balance. The latest biometric notice, effective May 2026, explains facial-geometry matching for avatar consent and anti-impersonation.