Magic AI Tops Deloitte's Fast 50 at 19,774%: Britain's Fastest-Growing Tech Company Is a Fitness Mirror

The Number That Stops the Room

Nineteen thousand, seven hundred and seventy-four percent. Three-year revenue growth. That's the figure that put Magic AI at number one in Deloitte's 2026 UK Technology Fast 50, announced Wednesday evening in London. Not a B2B SaaS platform. Not an LLM wrapper with enterprise pricing. A smart fitness mirror founded in 2022 by Varun Bhanot, out of London, that uses computer vision to watch you squat and tell you your knees are caving in.

I've been to enough of these awards dinners to know that the top slot usually goes to something that feels inevitable in retrospect. Monitise. Zoopla. Skimlinks in its heyday. This one felt different. Magic AI isn't just growing fast inside a niche. It's making a credible argument that the home fitness hardware market, which most people wrote off as a Peloton-shaped cautionary tale, has a second chapter.

What the Product Actually Is

Magic AI makes a wall-mounted smart display, cameras included, that merges real-time body tracking with coaching content and what they're calling ReflectAI, their proprietary computer vision layer for form correction and rep counting. Think less "Peloton with better branding" and more "the mirror as your most honest training partner." The one that doesn't let you skip the last set.

Users have completed more than 500,000 AI-coached workouts on the platform, logging over 150,000 hours of training data. That data flywheel is the real asset. Every workout makes the model marginally smarter. After enough sessions, ReflectAI stops being a form-checker and starts being a progression tracker, something that adapts its recommendations over time rather than simply clocking your reps.

On pricing: in the UK, there's a free Core tier included with the mirror, with a time-limited free-for-life class-library promotion for qualifying purchases. The optional MAGIC+ subscription runs £25 monthly or £199 annually, working out to roughly £16.58 a month. That's cheaper than a single personal training session in Manchester, let alone London.

I built something adjacent to this in 2007. A mobile wellness platform for corporate gyms. Cost me sixty grand, ran on J2ME, and had the engagement metrics of a government PDF. The tech wasn't ready. The distribution wasn't ready. The consumer appetite wasn't ready. What Magic AI has that we didn't: a smartphone-native audience completely unbothered by a camera watching them work out, a subscription model that creates recurring revenue from a hardware sale, and a dataset that compounds.

The Growth Story Behind the Growth Story

The Deloitte Fast 50 win doesn't exist in isolation. In September, Magic AI closed an £8 million Series B co-led by transatlantic investor Beringea and IW Capital, with participation from PXN Group, Accelerate Ventures, Active Partners, and Ventures Together. Total funding now sits at $20 million. The round was described as the largest AI fitness funding round in Europe this year, a fair claim for a category that most European VCs still treat as a curiosity rather than a platform.

The US expansion is the real bet. Magic AI already has over 18,000 members across all 50 states, pointing at a $35 billion AI fitness market. The Series B hires include executives with Peloton experience, which is either hard-won operational knowledge or a walking case study in what not to do with churn. Probably both. Whether that's a feature or a warning depends entirely on which part of Peloton's story those executives lived through.

Product-Market Fit: The Honest Assessment

Here's where I do the thing nobody does in awards coverage. I ask whether this is real.

Trustpilot reviews from UK buyers in 2026 are overwhelmingly positive. The hardware is praised for seamless integration into the home. The form-tracking gets called out specifically as genuinely useful rather than gimmicky. One reviewer from the Scottish Highlands bought the mirror because they lived twenty miles from a pool and ten from the nearest gym. That is product-market fit. Not a use case dreamed up by a product manager in Shoreditch. A real person solving a real problem.

The criticism that surfaces: delivery times occasionally disappoint, and the adjustable dumbbells Magic AI sells alongside the mirror have had quality complaints. The core product, the mirror itself, holds its rating. That's the right way round.

The subscription model is where this gets interesting. The 2026 Fast 50 cohort collectively generated £727 million in revenues with an average three-year growth rate of 2,824%. Magic AI is the outlier at the top. That growth rate isn't explainable by the hardware alone. Once a mirror is on the wall, the recurring subscription revenue is exceptionally sticky. The mirror is the acquisition cost. The subscription is the business.

The Trademark Signal

One data point worth sitting with: our analysis of IPO trademark filings shows Class 42 (software-as-a-service, technology services) UK applications at 3,748 in Q4 2026, down 54.2% versus the prior period. A dramatic contraction in SaaS trademark activity. Companies are either moving faster to market without the IP groundwork, or the pipeline of new SaaS ventures is genuinely thinning out.

Magic AI sits in a different category, hardware-led with a software layer, but the broader signal matters. The easy SaaS land-grab of 2021 to 2024 is over. The companies topping growth rankings in 2026 are the ones that solved a physical problem with a digital model. Peloton proved the category existed. Magic AI is making the argument that a leaner, smarter, British version of that thesis can outrun the American original.

So. A four-year-old London startup, founded in the year of the Bitcoin crash below £15,000, is now the fastest-growing technology company in the UK by revenue. The product watches you exercise and tells you off when you're doing it wrong. The dataset it's building is the kind of proprietary moat that makes acquirers get interested at exactly the wrong moment for founders.

Founder's Verdict

Magic AI is the real thing. The growth rate is extraordinary but the product mechanics justify the retention. Hardware plus subscription plus proprietary data is a durable model if, and only if, the hardware stays best-in-class and the content library keeps pace with consumer expectations. The US expansion is the right call. The risk is the same one that killed Peloton: that the American market will decide it wants to be sold fitness differently next year.

But right now, in November 2026, Varun Bhanot has built something that actually works. In a year of AI-flavoured vaporware and enterprise chatbots that charge £40,000 a year to summarise emails, that is worth saying out loud.

Fastest-growing tech company in the UK. And it's a mirror.