Sanctuary AI pricing

About

Sanctuary AI — legal name Sanctuary Cognitive Systems Corporation — is a Vancouver, British Columbia robotics company founded in 2018 by Geordie Rose, a serial deep-tech founder who previously co-founded quantum-computing pioneer D-Wave and robotics company Kindred. Its mission, in the company’s own words, is “to create and deploy millions of industrial-grade humanoid robots to help address growing labor challenges.”

The product is twofold: Phoenix, a general-purpose bipedal humanoid robot aimed at automotive, manufacturing, logistics, and retail work; and Carbon, Sanctuary’s proprietary AI control system that drives it. Sanctuary’s approach is teleoperation-first — human “pilots” remotely operate Phoenix to perform real tasks, capturing high-fidelity behavioural and tactile data that is used to train Carbon toward autonomy. Phoenix is now on its eighth generation (unveiled January 2025), each iteration optimized for data capture, dexterity, and lower manufacturing cost.

The company has raised over $140 million in total (as of July 2024), including a C$75.5 million (US$58.5 million) Series A (2023, with Bell, Evok Innovations, Export Development Canada, Magna, SE Health, Verizon Ventures and Workday Ventures), a $30 million grant from Canada’s Strategic Innovation Fund (2022), and later strategic financing from BDC Capital and InBC. Tellingly, several of its biggest names — Magna (a multi-unit Phoenix deployment in automotive sub-assembly, plus a potential manufacturing partner), Accenture (channel access to Global 2000 clients via Project Spotlight), and Microsoft (AI R&D) — are simultaneously investors, customers, and distribution.

Crucially for a pricing blueprint: Sanctuary AI publishes no pricing of any kind. The website (sanctuary.ai) exposes only Technology, About, Careers, and News, with the only commercial call-to-action being “Contact Us”; sanctuary.ai/pricing returns a 404. There is no plan grid, no per-unit price, no robot-as-a-service hourly rate, and no subscription. So this page documents what is honestly known — the company, its funding, its partners, its teleoperation model, and its sales-led posture — rather than inventing a number Sanctuary has never published.

Pricing summary : a sales-only humanoid-robot lab with no published rate

Sanctuary AI runs a sales-led, no-public-price commercial model. There is no subscription, no published per-unit price, no robot-as-a-service (RaaS) hourly rate, and no self-serve tier to evaluate. The dimensions, such as they can be observed, are:

• Enterprise pilots & deployments — quoted bespoke via “Contact Us”. No floor price, packaging, or unit cost is disclosed. This is the company’s revenue surface: Phoenix humanoids deployed into real operations (e.g. Magna’s automotive sub-assembly), scoped and priced per customer and per deployment.
• Teleoperation-as-a-service, in effect — because Phoenix is operated by human pilots that feed data to Carbon, what an early customer is really buying is remotely-operated labour that improves over time, not a fully autonomous product with a clean per-unit price. The value metric is implicit (work performed) rather than a published meter.
• No free tier, no trial — Phoenix is physical hardware deployed under commercial agreement; there is nothing to sign up for or download.

What makes this different: unlike humanoid-RaaS peer Agility Robotics — which has publicly framed its Digit robot as a per-hour subscription — Sanctuary AI discloses no commercial rate at all. And unlike a metered software lab such as Essential AI, there are not even free artifacts on the side: the entire offering, hardware and software, sits behind “Contact Us”. The only numbers ever attached to a Phoenix unit are third-party estimates on aggregator sites — not Sanctuary’s own price.

Pricing by product

Sales motions across products: fully sales-led. The only revenue-bearing surface (Phoenix deployments) is quoted via “Contact Us”. There is no self-serve motion, no free tier, and no published meter — the data above reflects the absence of public pricing, not specific figures Sanctuary has released.

Hidden costs : What Sanctuary AI customers actually pay

Because Sanctuary AI publishes no price, the “real bill” question can’t be answered from public data — but the shape of the cost is unusually clear for a humanoid deployment, and it is more than a hardware sticker.

The unit is only the start. Even using the third-party estimates (around USD 65k “from”; USD 100k–250k per-unit ranges that circulate on aggregator sites — again, not Sanctuary figures), a Phoenix deployment carries costs a software buyer never sees: integration into a physical line, safety and workspace adaptation, ongoing maintenance, and — critically — teleoperation labour. Because Phoenix is piloted by humans during much of early deployment, the customer is effectively paying for remotely-operated work, so a meaningful share of the real cost is the human-in-the-loop operation, not just the machine.

Evaluation friction is total. With no published floor, no packaging, and no worked example, a prospective buyer cannot estimate cost without a sales conversation and a scoped pilot. There is no calculator a buyer can self-serve against, and no RaaS rate to multiply by hours.

Want to model what a humanoid-robot or robot-as-a-service deployment might cost? There’s no published Sanctuary rate to plug in, but you can sketch scenarios with the Sanctuary AI pricing calculator, and for how to choose a meter when the deliverable is work performed rather than a clean unit, see choosing the right usage metric.

Pricing evolution : Sanctuary AI pricing history and changes

Sanctuary AI has never had a public price to change. Its “pricing evolution” is really a commercial-posture evolution: funded humanoid lab, then strategic-investor-as-customer pilots (Magna, Accenture, Microsoft), then a maturing teleoperation-to-autonomy data flywheel — all while keeping its actual commercial offering behind “Contact Us.” The milestones below are reconstructed from primary announcements and a live 2026-06-14 site check.

Cadence

Tracked range: 2023 Q1–2026 Q2. Zero public price changes across the company’s life — there has never been a published price to revise. Quarters not listed had no relevant public pricing event.

Notable changes

• 2023-03 — Closes a C$75.5M (US$58.5M) Series A (Bell, Evok, Export Development Canada, Magna, SE Health, Verizon Ventures, Workday Ventures) to accelerate Phoenix + Carbon. Sales-led; no price card (sanctuary.ai blog).
• 2024-03 — Accenture Ventures invests (amount undisclosed); Sanctuary gains Project Spotlight access to Global 2000 clients; Phoenix had by then run 110+ tasks at a Mark’s retail store in Langley, BC (Accenture newsroom).
• 2024-07 — Strategic financing from BDC Capital + InBC brings total investment over $140M (PR Newswire).
• 2025-01 — Phoenix Generation 8 unveiled, optimized for data capture plus new tactile sensors — advancing the teleoperation-to-autonomy flywheel (The Robot Report).
• 2026-06-14 — Live check confirms no public pricing: sanctuary.ai/pricing 404s; the only commercial CTA is “Contact Us”.

What’s unique : Sanctuary AI’s distinctive pricing mechanics

1. The price is the absence of a price. Where a humanoid-RaaS peer like Agility Robotics has anchored on a public per-hour rate, Sanctuary AI publishes nothing commercial — no unit price, no RaaS rate, no floor, no calculator. The entire monetized surface (Phoenix deployments) sits behind a single “Contact Us” CTA. For a company selling bespoke, integration-heavy humanoid pilots, opacity is the deliberate packaging: every deployment is scoped and quoted, so there is no list price to commoditize against.

2. Investors are the customers — and the channel. Magna, Accenture, and Microsoft are simultaneously cap-table backers, deployment sites, and distribution. That tight loop means early “pricing” is really negotiated inside strategic partnerships rather than offered on a rate card — a Phoenix unit at Magna is part of a broader relationship, not a transaction with a sticker. It’s a go-to-market that substitutes partnership for pricing.

3. You’re buying labour-as-a-service, not a product unit. Because Phoenix is teleoperation-first, an early buyer is really purchasing work performed by a human-piloted robot that improves toward autonomy — closer to staffing than to buying a machine. That makes a clean per-unit or per-token price genuinely hard to set: the deliverable is an outcome (tasks completed) whose autonomy ratio shifts over time. (For why outcome-shaped deliverables resist a public unit price, see outcome-based pricing trends.)

Strengths & weaknesses

Billing UX : Sanctuary AI billing controls and transparency

• Billing controls — None are public. There is no self-serve dashboard, no usage meter, and no plan-management UI exposed on the site; commercial terms are handled entirely through a sales relationship and bespoke pilot contracts.
• Usage visibility — Not applicable as a public product. Any operational telemetry (tasks completed, teleoperation hours, uptime) lives inside a deployed customer’s pilot, not in a public Sanctuary console or rate card.
• Payment options — Not disclosed. Enterprise deployments are presumably invoiced under custom contracts negotiated via “Contact Us”; there is no card checkout or public billing portal.
• Transparency — Low on the commercial side by design. Sanctuary is communicative about its technology (Phoenix generations, Carbon, tactile sensors, teleoperation) but discloses nothing about how a deployment is priced or metered.

Strategic wins : Why Sanctuary AI’s pricing decisions worked

1. Partnership instead of a price sheet

By turning Magna, Accenture, and Microsoft into investors and customers and channel, Sanctuary lands real deployments (e.g. automotive sub-assembly) without ever needing to publish a rate. The relationship carries the commercial terms; the price stays private. For an early, integration-heavy hardware product, this de-risks go-to-market far better than a public price card would — and it mirrors how other AI companies are rethinking what they meter.

2. No list price, no commoditization

For a company selling bespoke humanoid pilots, publishing a unit price or RaaS rate would invite immediate comparison against rivals on a single number. By quoting every deployment, Sanctuary keeps pricing tied to the integrated value delivered and protects its negotiating position — a defensible stance when the offering is a custom, labour-replacing deployment, not a fungible unit. See usage-based pricing strategy for when quoting the outcome beats metering the unit.

3. Sell work now, monetize autonomy later

The teleoperation-first model lets Sanctuary deploy Phoenix into paying pilots before full autonomy exists, while each deployment generates the very data that improves Carbon. That sequencing — get into operations early, improve the autonomy ratio over time — is hard to express as a clean per-unit price, so keeping pricing bespoke fits the product’s actual maturity curve. Related: outcome-based pricing trends.

Areas to improve : Gaps in Sanctuary AI’s pricing approach

1. Publish something — even a model, not a number

A fully dark price page (a literal 404) maximizes evaluation friction for anyone outside the strategic-partner circle. Even a one-line description of how deployments are priced (per unit? per teleoperation hour? per task / outcome?) would let prospective buyers self-qualify instead of bouncing. The total opacity invites exactly the cost-unpredictability anxiety that scares buyers off a major capital commitment.

2. Clarify the teleoperation cost story

The biggest hidden cost — that early Phoenix work is human-piloted — is invisible in any public framing. A transparent account of the autonomy ratio (how much is robot vs. remote operator today, and the trajectory) would help buyers understand what they’re really paying for, and would let Sanctuary frame a credible labour-cost comparison rather than leaving the whole economics to a cold sales call.

3. Consider a published RaaS anchor as the product matures

Humanoid peers have shown that a robot-as-a-service per-hour rate (Agility Robotics) gives buyers a comparable, budgetable unit. As Phoenix’s autonomy improves, a published RaaS anchor — even a “starting at” figure — would shorten evaluation cycles and let mid-market manufacturers (not just strategic partners) enter the funnel. Compare how other AI companies stage enterprise transparency.

Key takeaways

1. No public price is itself a pricing decision. Sanctuary AI publishes zero commercial pricing and routes everything through “Contact Us” — a deliberate, sales-led posture, not an oversight. For bespoke, integration-heavy humanoid pilots, opacity protects margin and negotiating leverage.
2. Partnership substitutes for a price sheet. Magna, Accenture, and Microsoft are investors, customers, and channel at once, so early “pricing” is negotiated inside relationships rather than offered on a rate card.
3. You’re buying labour, not just a machine. Phoenix is teleoperation-first, so a deployment is really human-piloted work that trains toward autonomy — a value metric that genuinely resists a clean per-unit price.
4. The only numbers are third-party guesses. Aggregator estimates (roughly USD 65k–250k per unit) are not Sanctuary figures; the company itself has never published a unit price or RaaS rate across eight Phoenix generations.
5. Total opacity is a double-edged sword. It shields Sanctuary from commoditization but maximizes buyer friction; the obvious improvement is a minimal public anchor (even a pricing model) without giving up the bespoke-quote approach.

UBP implications

1. Some deliverables are best priced by not publishing a price — yet. When the offering is a bespoke, integration-heavy humanoid deployment whose autonomy ratio is still changing, a public unit price can mislead more than it helps. UBP practitioners should match transparency to product maturity — quote bespoke outcomes while the unit is still moving, and publish a meter only once it stabilizes.
2. Partnership can be a pricing channel. Sanctuary shows that strategic-investor-customers can carry commercial terms that would otherwise need a public rate card. The lesson: when distribution and demand come bundled in a relationship, the “price” can live inside the partnership rather than on the website — at the cost of reach beyond those partners.
3. Teleoperation reframes the meter. A robot that is partly a remote human operator is closer to labour-as-a-service than to a hardware sale. UBP designers should be explicit about what is metered (the unit, the hours, or the task completed), because conflating a one-time machine price with ongoing human-in-the-loop cost erodes trust the moment the real bill arrives.

Sources

• Sanctuary AI — official site (no pricing page; “Contact Us” CTA) (accessed 2026-06-14)
• Sanctuary AI — Technology (Phoenix + Carbon) (accessed 2026-06-14)
• Sanctuary AI — News (accessed 2026-06-14)
• Sanctuary Cognitive Systems Closes C$75.5M (US$58.5M) Series A (sanctuary.ai) (accessed 2026-06-14)
• Accenture Invests in Sanctuary AI (Accenture newsroom, 2024-03-27) (accessed 2026-06-14)
• Sanctuary AI enters strategic relationship with Magna (The Robot Report) (accessed 2026-06-14)
• Sanctuary AI strategic financing from BDC Capital and InBC; total tops $140M (PR Newswire, 2024-07) (accessed 2026-06-14)
• Sanctuary AI integrates tactile sensors into Phoenix (The Robot Report, 2025) (accessed 2026-06-14)
• Microsoft taps Sanctuary AI for general-purpose robot research (TechCrunch, 2024-05-01) (accessed 2026-06-14)
• Browse the pricing blueprint corpus

Bottom line

Sanctuary AI is a clear sales-only case in the embodied-AI corpus: a Vancouver humanoid-robot company run by serial deep-tech founder Geordie Rose, $140M+ raised (including a $30M Canadian government grant), that publishes no price for anything. Its commercial offering — Phoenix humanoids driven by the Carbon AI control system — sits entirely behind “Contact Us”; sanctuary.ai/pricing is a literal 404. Sanctuary’s go-to-market substitutes partnership for pricing: Magna, Accenture, and Microsoft are investors, customers, and channel at once, so deployments are negotiated inside relationships rather than offered on a rate card. And because Phoenix is teleoperation-first, an early buyer is really purchasing human-piloted labour that trains toward autonomy — an outcome-shaped deliverable that genuinely resists a clean public unit price. The only numbers attached to a Phoenix unit are third-party estimates, never Sanctuary’s own.

Want to compare Sanctuary AI against a humanoid-robot peer that has talked pricing? See Agility Robotics, or browse the full pricing blueprint.