How Synthesia Scaled AI Avatar Business Video

Why Synthesia matters in 2026

Synthesia is the British AI video startup that turned a research demo into the operating system for corporate talking heads. As of January 2026, the London-based company is valued at $4 billion after a $200 million Series E led by Alphabet’s GV (The Guardian, 26 Jan 2026; Forbes, 29 Oct 2025). It claims 70% of the FTSE 100 and more than 90% of the Fortune 100 as customers (The Guardian, 26 Jan 2026; Synthesia case studies).

That’s a wild trajectory for a company that began as a deepfake David Beckham malaria-awareness stunt in 2019. We have been tracking Synthesia since the early days, and what we see in mid-2026 is no longer a clever demo. It is a default tool inside the learning, enablement, and internal-comms workflows of the world’s largest companies.

What Synthesia actually sells

Quick answer: Synthesia is a browser-based AI video studio that turns a typed script into a presenter-led video using one of 240+ stock avatars, a personal avatar you upload, or a custom Studio avatar. It ships with 1,000+ AI voices across 160+ languages and includes 1-click translation, AI dubbing with lip-sync, brand kits, interactive video tools, an API, and an LMS-friendly SCORM export.

The product has shifted from a single feature (text-to-avatar) into a full platform with five jobs to do:

1. Create – type a script, pick an avatar, render a video.
2. Localize – translate and dub finished videos into 140+ languages with synced lips.
3. Manage – brand kits, live collaboration, workspaces, version control.
4. Publish – shareable video pages, SSO gating, analytics, embed anywhere.
5. Engage – quizzes, branching, clickable CTAs, and the new “Roleplay Sessions” for sales coaching.

That last bucket is where Synthesia is now pushing hardest. In July 2026, the company is promoting a “new release” called Roleplay Sessions, which lets learners practice real conversations with interactive AI avatars and get scored feedback (Synthesia pricing page).

“We don’t care about AI video as in ‘here’s a thing that can make absolutely anything you can think of.’ We only care about humans and videos and presenters for business content.” — Victor Riparbelli, Synthesia CEO, in Forbes (29 Oct 2025)

The founders and the long view

Synthesia was co-founded in 2017 by four people with serious computer-vision credentials: CEO Victor Riparbelli and COO Steffen Tjerrild (both Danish), plus UCL computer-vision professor Lourdes Agapito and TUM professor Matthias Niessner (Wikipedia – Synthesia (company); Synthesia about page). The founding team came out of Stanford, Cambridge, UCL, and TUM.

The first public proof point was a 2018 BBC Click segment showing a digital clone of newsreader Matthew Amroliwala speaking Spanish, Mandarin, and Hindi (BBC). The team then built an early business on synthetic celebrity translations and a 2021 Lay’s “Messi Messages” campaign that let fans send personalized AI video from Lionel Messi — a Bronze Cannes Lion winner (Cannes Lions entry).

But the company nearly ran out of runway in its first two years. In a 2019 TechCrunch profile, founder Riparbelli said Synthesia had only ten employees and was struggling to make sales, which forced a pivot from entertainment clients to enterprise training (TechCrunch, 2019; cited in Wikipedia). That pivot is the reason a 2026 customer list reads like a corporate phone book.

Funding, revenue, and the Adobe that almost was

Synthesia’s funding history tracks the generative-AI boom and the 2023–2026 enterprise-AI correction that followed it.

• 2019: $3.1M seed, led by LDV Capital.
• April 2021: $12.5M Series A led by FirstMark.
• December 2021: $50M Series B led by Kleiner Perkins and GV (Reuters, 8 Dec 2021).
• June 2023: $90M Series C led by Accel and NVentures, taking Synthesia to a $1 billion valuation and unicorn status (BusinessWire; TechCrunch, 14 Jun 2023).
• January 2025: $180M Series D led by NEA, doubling the valuation to $2.1 billion (Reuters, 15 Jan 2025; CNBC, 15 Jan 2025).
• April 2025: Adobe took an undisclosed stake in a “strategic” partnership, valuing Synthesia around $2.1B; the company also passed $100M in annual recurring revenue (CNBC, 15 Apr 2025).
• October 2025: Reported $200M round led by GV at a $4 billion valuation; Synthesia rejected a $3 billion acquisition offer from Adobe earlier in the year (Forbes, 29 Oct 2025; The Times).
• January 2026: GV-led round officially closed at $4B post-money (The Guardian, 26 Jan 2026).

On revenue, The Guardian reported Synthesia’s 2024 revenue at $58.3M with a pre-tax loss of $59.2M, with the company on track for $200M in revenue in 2026 (The Guardian, 26 Jan 2026). Forbes had earlier reported Synthesia crossed $100M in annualized revenue in April 2025 (Forbes). The headcount grew from 400 people in 2025 to about 600 by January 2026 (CNBC, 15 Jan 2025; The Guardian, 26 Jan 2026).

Who actually uses Synthesia in 2026

The customer logo wall reads like a global index. The company’s own case studies and the Wikipedia entry document deployments at Reuters, SAP, Zoom, Heineken, Moody’s, Merck KGaA, Mondelez, Carlsberg, KONE, ERGO, The Siegfried Group, Young Life, Tiffany & Co., IHG Hotels & Resorts, the BBC, Uber, and Zoom (Synthesia case studies; Wikipedia). The Guardian adds NatWest, Lloyds Bank, British Gas, the NHS, the European Commission, and the United Nations to the UK public-sector and enterprise list (The Guardian, 26 Jan 2026). Forbes names DuPont, Xerox, and Spirit Airlines as US enterprise customers using Synthesia for safety briefings and other training (Forbes, 29 Oct 2025).

If you are sizing the market for AI video, the simpler read is this: Synthesia is now embedded inside L&D and enablement stacks the same way Salesforce is embedded in sales pipelines. Three numbers we trust:

1. 60,000+ customers as of January 2025 (Reuters).
2. 1M+ users and 20M+ avatar videos generated as of 2026 (Synthesia pricing; Synthesia avatars).
3. 150,000 personal avatars created by 65,000 customers as of June 2025 (Fortune, 26 Jun 2025).

Pricing and packaging in July 2026

Synthesia’s pricing has come down over the last 18 months. The current page advertises “plans now starting from $18/month — save 38%” as a limited-time cut, and the published tiers we verified in July 2026 are:

Plan	Price (billed monthly)	Video minutes	Key limits	Personal Avatars
Basic	$0	10 min / month	9 stock avatars, Synthesia watermark	—
Starter	$29/mo	10 min / month	125+ stock avatars, downloads, AI Video Assistant, AI Dubbing, 1 editor + 3 guests	3
Creator	$89/mo	30 min / month	180+ stock avatars, multiple avatars per scene, interactive video, API, 1 editor + 5 guests	5
Enterprise	Custom	Unlimited	240+ stock avatars, unlimited personal avatars, 1-click translation into 80+ languages, SAML/SSO, SCORM export, Brand Kits, dedicated CSM	Unlimited

Source: Synthesia pricing page (live as of 28 July 2026). Credit system now uses a single “credits” currency across all AI features, with 1,200 credits/mo on Basic and Starter, 3,600 credits/mo on Creator, and custom credit pools on Enterprise. Synthesia is also running a July–August 2026 promotion giving every user 15 free dubbing minutes per day (Synthesia pricing).

Synthesia vs HeyGen, Runway, D-ID, and Hour One

Synthesia does not own the AI video market. The honest comparison in 2026 is that each major tool occupies a different lane. We pulled current public pricing for the closest peers in July 2026.

Tool	Best lane	Starting price (verified)	Key limits	Notable differentiator
Synthesia	Business avatar presenters, training, L&D	$0 / $29 / $89 / custom	Minute-based, 240+ stock avatars	Largest enterprise L&D customer base, ISO 42001, 1M+ users
HeyGen	Marketer/creator avatar video and translation	$0 / $29 / $49 / $149 / custom	175 languages, 700+ stock avatars on Business	1,000+ voice library, Avatar IV/V engines, 4K export on Pro
Runway	Generative cinematic video and VFX	$0 / $12 / $28 / $76 / custom	Credit-based across Gen-4.5, Aleph 2.0, Seedance 2.0	Open-ended generative video, not avatar-locked
D-ID	Talking-head video from a photo, API	Free trial + paid Studio plans	500+ stock avatars, V4 Expressive	Real-time self-hosted agents, photo-to-video specialty
Hour One	AI news-style presenters and enterprise video	Tiered, not public	Studio-grade virtual humans	Used in news and corporate communications

Sources: Synthesia pricing; HeyGen pricing; Runway pricing; D-ID pricing.

The honest read is that Runway is the right tool for the cinematic ad and the abstract b-roll piece. HeyGen is a strong direct competitor for the avatar use case, with a heavier creator-marketing tilt and a lower per-seat entry price ($29/mo on Creator, the same as Synthesia’s Starter). D-ID and Hour One occupy niches — D-ID on the photo-to-video API, Hour One on news-style presenters. None of them, as of July 2026, has Synthesia’s combination of enterprise L&D integrations, SCORM export, and Fortune-100 install base.

Latest 2025–2026 product launches

Synthesia’s roadmap in the last 18 months has leaned hard into agentic AI, expressive avatars, and tighter creative control. The launches we verified:

• Expressive Avatars (April 2024): Text-driven micro-expressions, body language, voice tone. The launch that put “Synthesia 2.0” on the map (CNBC, 25 Apr 2024).
• Express-2 model (August 2025): A new rendering model with parameters “in the billions” that more accurately captures facial movement, hand gestures, and the speaker’s natural accent (MIT Technology Review, 4 Sep 2025).
• Veo 3 integration (2025): Direct generation of Google DeepMind’s Veo 3 cinematic b-roll inside Synthesia scenes, available on Customizable Avatars (MIT Technology Review; Synthesia avatars).
• AI Dubbing 2.0 (July 2026): Translate and re-lip any video — not just Synthesia-made ones — into 140+ languages, with voice and accent preservation (Synthesia pricing).
• Synthesia 3.0 (October 2025): Hyper-realistic avatars with built-in AI dubbing, AI Video Assistant, and a new agentic layer for talking-head “video agents” (Bloomberg, 1 Oct 2025; MIT Technology Review).
• Roleplay Sessions (2026): New release. Learners hold simulated conversations with an avatar and get scored feedback (Synthesia pricing).
• Interactive Avatars (Beta, 2026): Avatars that listen, talk, and respond in real time inside the customer’s own product; customers can connect their own LLM or agent to the avatar layer (Synthesia avatars).

The 2026 direction is clear: Synthesia is moving from one-way talking-head video into conversational, agent-driven training. As MIT Technology Review noted, future versions could let a learner ask the avatar to pause, expand, or answer a follow-up — something competitors like HeyGen and D-ID are also racing toward.

Integrations: where Synthesia plugs in

A list of every integration would be a small book. The high-value ones we verified on the public docs page (Synthesia integrations) are:

• LMS: SAP Litmos, Docebo, TalentLMS, Moodle, 360Learning, aNewSpring, Articulate 360, Lectora, Thinkific, Thought Industries, Udemy, Continu, Coassemble, Easygenerator, eloomi, Eduflow, Kumullus, ParkOOroo, Stellar Labs, uQualio, Wisetail, Mini Course Generator, ETU.
• HRIS / workforce: Workday, BambooHR, HiBob.
• Marketing / support / CMS: HubSpot, Intercom, Squarespace, WordPress, Shopify, BigCommerce, Kickstarter, Medium, Notion, Vimeo, YouTube, Mindstamp, Tolstoy, Userpilot, ChurnZero, Zngly, Crowdin, Blanc, FlowShare, Excel Add-in.
• Productivity and automation: Slack, Microsoft Teams, Monday.com, Make, Workato.
• Authoring utilities: PowerPoint, Google Drive, Unsplash, Excel.

The reach matters because Synthesia’s enterprise pitch is not “we render video.” It is “your video can be SCORM-exported into your LMS, gated by your SSO, and updated globally from one script.” That is the workflow Runway, HeyGen, and D-ID still trail on.

Security, ethics, and the deepfake problem

Synthesia’s biggest public risk has always been the same risk that built the company: someone using its avatars to impersonate a real person without consent. The company has been public about its guardrails since 2018, and the 2026 stack is more thorough than most peers.

The verified controls in July 2026:

• SOC 2 Type II, ISO 27001, ISO 27701, and ISO 42001 certified — Synthesia was the first generative AI video company certified to ISO 42001 (Synthesia responsible AI; Synthesia avatars).
• GDPR compliant and EU AI Act Article 50 labeled on all generated content (Synthesia about).
• Live consent verification for every personal avatar — the person must record a live, on-camera consent clip that matches the avatar’s likeness, and the consent clip cannot be uploaded (Synthesia avatars).
• Content moderation at the point of creation, with both human and AI review (Synthesia responsible AI).
• AI Futures Council of independent experts, including deepfake researcher Henry Ajder, journalist Sophia Smith Galer, AI policy expert Dani Dhiman, and AI-education specialist Dr. Phillipa Hardman (Synthesia responsible AI).
• Talent Experience Program that gives the human actors behind stock avatars a share of company stock, as reported by the Financial Times in March 2025 (FT, 25 Mar 2025).

This is the right scaffolding, but the company has not been incident-free. An October 2023 Freedom House report found that Synthesia’s tools were used by state-aligned actors in Venezuela, China, Burkina Faso, and Russia to produce fake TV news broadcasts. The Guardian and Deadline both reported on British actor Dan Dewhirst, who signed with Synthesia in 2021 and later found his avatar used in Venezuelan propaganda (The Guardian, 16 Oct 2024; Deadline, 17 Oct 2024). Synthesia told the New York Times in February 2024 that it had tightened misuse detection and that all new users are screened by a moderation team (NYT, 16 Feb 2024). The company still explicitly bans political and news-like content on stock avatars, per its own responsible AI page.

Real limitations we have seen

Synthesia is the best-in-class for what it does, but it is not a general-purpose video tool. The constraints we would tell a buyer about:

1. Studio-true aesthetic is still beyond reach. Even the Express-2 model in 2026 has telltale signs of AI — bright pink palms, stiff hair, glassy eyes, occasional audio drift (MIT Technology Review, 4 Sep 2025). Use it for internal training, not for hero brand film.
2. Minutes and credits cap lower tiers. The $29 Starter plan gives you 10 video minutes and 10 dubbing minutes per month. The $89 Creator plan gives you 30 of each. High-volume teams hit those limits fast (Synthesia pricing).
3. Studio Avatars cost extra. Synthesia’s “Studio Express-1” custom avatar is a $1,000/year add-on for annual customers, and it takes up to 10 days to process (Synthesia pricing).
4. Compliance review is on you. Synthesia moderates for political, sexual, and extremist content, but factual accuracy and brand-legal review are still on the customer. A typo in your script becomes a typo on the avatar’s face.
5. The deepfake floor is non-zero. No moderation system catches every misuse. If you operate in a regulated or news-adjacent industry, plan for the residual risk.

Founder and operator quotes worth remembering

We pulled the lines that have shaped the public Synthesia story, because they are useful context for any team weighing the platform.

• Victor Riparbelli on focus: “We don’t care about AI video as in ‘here’s a thing that can make absolutely anything you can think of.’ We only care about humans and videos and presenters for business content.” — Forbes, 29 Oct 2025.
• Steffen Tjerrild on the round: “Existing investors have seen the progress, have seen the numbers, have seen them compound year over year. This round is led by insiders, or predominantly existing investors that deeply understand the business.” — The Guardian, 26 Jan 2026.
• Tjerrild on AI and jobs: “We’re an AI-first company, we have 600 employees and we hired 40% more people last year.” — The Guardian, 26 Jan 2026.
• Peter Hill (CTO) on Veo 3: Confirmed in 2025 that Synthesia will integrate Google DeepMind’s Veo 3 to generate cinematic b-roll inside avatar scenes (MIT Technology Review, 4 Sep 2025).
• Youssef Alami Mejjati (Head of R&D) on the rendering jump: “With Express-1, it needed to first see someone expressing emotions to be able to render them. Now, because we’ve trained it on much more diverse data and much larger data sets … it just learns these associations automatically.” — MIT Technology Review, 4 Sep 2025.
• Frank van der Grijspaarde (Heineken, Global TPM Manager) on adoption: “I’ve never even used a tutorial, because the platform is so easy to use. This actually is the first time in my career that I’ve been an early adopter of new technology.” — Synthesia Heineken case study.
• Alex Voica (Head of Corporate Affairs) on Zoom’s savings: “Zoom achieved monthly cost savings of $1,000–$1,500 per employee which was previously spent on creating training videos.” — VentureBeat, 31 Jan 2024.

FAQ: what teams ask us about Synthesia

What is Synthesia used for?

Synthesia is used to make presenter-style business videos — training, internal comms, sales enablement, marketing explainers, and localization — without filming a human. The company’s own case studies show deployments at Heineken, SAP, Merck, Mondelez, Carlsberg, KONE, and ERGO for these jobs (Synthesia case studies).

How much does Synthesia cost?

In July 2026, Synthesia’s published pricing starts free on the Basic plan, $29/mo on Starter (10 video minutes), $89/mo on Creator (30 video minutes), and custom on Enterprise (unlimited minutes, SAML/SSO, SCORM, dedicated CSM). The “save 38%” promo on the current page puts the floor at $18/mo annual equivalent (Synthesia pricing).

Is Synthesia SOC 2 compliant?

Yes. Synthesia is SOC 2 Type II, ISO 27001, ISO 27701, and ISO 42001 certified, and is GDPR compliant. It was the first generative AI video company certified to ISO 42001 (Synthesia responsible AI).

How does Synthesia compare to HeyGen?

Both are avatar-led. HeyGen starts at $29/mo on Creator with a heavier creator-marketing tilt, 700+ stock avatars on Business, and 175 languages (HeyGen pricing). Synthesia’s edge is the enterprise L&D install base (70% FTSE 100, 90% Fortune 100), the LMS/SCORM integrations, and the larger voice library (1,000+) (Synthesia pricing). For one creator making social clips, HeyGen is a great pick. For an L&D team running global programs, Synthesia still wins.

What about deepfakes and ethics?

Synthesia requires live on-camera consent for every personal avatar, runs both human and AI moderation, is a Content Authenticity Initiative member, and explicitly bans political and news-like content. There have been documented abuse cases (see the 2023 Freedom House report and the 2024 Guardian/Deadline reporting on Dan Dewhirst), which is why every customer should still run its own legal review on high-risk scripts (Synthesia responsible AI; Freedom House via Gizmodo; The Guardian, 16 Oct 2024).

Can I use Synthesia in my LMS?

Yes. Synthesia exports SCORM packages and has direct integrations with SAP Litmos, Docebo, TalentLMS, Moodle, Articulate 360, and 20+ other LMS and LXP platforms (Synthesia integrations).

Does Synthesia do AI dubbing of my existing videos?

Yes. AI Dubbing 2.0 launched in July 2026 and translates any uploaded video (not just Synthesia-made ones) into 140+ languages with voice and accent preservation, with optional lip-sync that doubles credit usage (Synthesia pricing).

How many languages does Synthesia support?

Synthesia supports 160+ languages and accents for the in-platform avatar videos and 140+ for AI Dubbing 2.0. Enterprise customers get 1-click translation into 80+ languages (Synthesia pricing; Synthesia avatars).

Is Synthesia worth it for a small business?

If you produce fewer than 10 training or marketing videos per quarter, the Basic or Starter plan is the right starting point. If you need unlimited personal avatars, SCORM export, and SSO, the Enterprise plan is where the operational savings show up. The Zoom case study — $1,000–$1,500 per employee per year saved on training video production — is the most defensible ROI benchmark (VentureBeat, 31 Jan 2024).

Is Synthesia profitable?

No, and that is not the focus. Synthesia reported a pre-tax loss of $59.2M against $58.3M in revenue in 2024, and CEO Riparbelli told CNBC in April 2025 that profitability is “not an immediate focus” (The Guardian, 26 Jan 2026; CNBC, 15 Apr 2025).