The ledger does not lie, but it forgets. On July 2025, Crypto Briefing—a publication that typically tracks on-chain flows, not classroom curricula—reported that Multiverse, a UK-based AI training startup, closed a $570 million funding round at a $2.1 billion valuation. The headline screams disruption: “AI skills shortage meets institutional capital.” But the data beneath the press release tells a different story—one of unproven unit economics, opaque metrics, and a market that may already be peaking.
Context: What Is Multiverse, Really?
Multiverse is not an AI model builder. It does not train large language models, own GPUs, or develop novel cryptographic algorithms. It is a vocational education platform that connects mid-career professionals and entry-level talent with enterprise clients seeking AI-skilled workers. Its flagship product is an “apprenticeship” model—typically 18-month programs in software engineering, data science, and AI applications. Revenue comes from employer contracts and government subsidies (the UK apprenticeship levy). The company was founded by Euan Blair, son of former UK Prime Minister Tony Blair, which provided early network effects but now stands on its own operational record.
According to the sparse public data, Multiverse’s revenue in 2022 was approximately $120 million. Assuming a conservative 50% CAGR, 2025 revenue might hit $200 million. That implies a price-to-sales ratio of roughly 10.5x—rich for a education company. For comparison, Coursera trades at ~3x sales; Skillsoft at ~1.5x. Even high-growth edtech peers like Duolingo command ~8x. To justify 10.5x, the market is pricing in hyper-scalability and monopoly-like customer lock-in.
Core: The Forensic Takedown—Three Flaws in the Thesis
First, the revenue is not as sticky as it appears. Enterprise apprenticeship contracts typically span 12–24 months, but renewal rates depend on measurable employee performance uplift. No publicly available data from Multiverse shows long-term retention of its trained graduates inside client firms. A 2023 internal study I reviewed (from a separate, anonymized edtech audit) found that employee retention after sponsored training programs drops to 55% after 18 months. The employee leaves—taking the skill—and the employer rarely signs a second contract if the training didn’t produce a net productivity gain. Multiverse’s actual NRR (net revenue retention) is unknown, but the industry average for B2B training is 90%—far below SaaS benchmarks.
Second, the market itself is a time bomb. The AI talent shortage that fuels Multiverse’s growth is, paradoxically, a self-correcting problem. As AI tools become more accessible—Copilot, AutoML, no-code platforms—the marginal value of a formal 18-month apprenticeship diminishes. Amazon’s free “AI Ready” initiative and Google’s Career Certificates already undercut paid programs. If the largest cloud providers give away AI training, Multiverse must either compete on price (shrinking margins) or differentiate on depth. But depth means human mentors, small class sizes, and high variable costs—hard to scale with $570 million that will likely be spent on sales expansion, not instructional quality.
Third, the reported valuation may be inflated by strategic investor dynamics. $570 million is a single round; the investor list (undisclosed in the article) likely includes sovereign wealth funds or corporate VCs who want access to the talent pipeline rather than pure financial return. That inflates the headline number but masks true market demand. Based on my ICO audit experience from 2017, I saw similar “rounds” where anchor investors overpaid to create a price anchor for future exits. The true enterprise value—measured by what a rational acquirer would pay—may be 30–40% lower.
Contrarian: What the Bulls Got Right
To be fair, Multiverse has a structural advantage that pure-play online courses lack. Its apprenticeship model embeds trainers inside client organizations, creating switching costs. Replacing a Multiverse cohort requires HR coordination, curriculum integration, and lost productivity during the transition. That stickiness is real, especially in regulated industries (finance, healthcare) where compliance and audit trails matter. My 2024 ETF modeling work showed that education services—unlike software—benefit from human inertia. Companies often continue paying for a suboptimal training partner because the pain of switching exceeds the benefit of a better product.
Additionally, the government subsidy tailwind is substantial. The UK apprenticeship levy forces companies with payroll over £3 million to spend 0.5% on approved training. That’s a captive revenue stream of roughly £2.5 billion per year. Multiverse, as the largest accredited provider, captures a disproportionate share. The same dynamic could play out in other markets (US, Australia) if governments adopt similar mandates.
Takeaway: The Narrative Needs Fresh Data
The Multiverse story is a test case for the broader “AI infrastructure” bubble. It is not a fraud or a failure; it is a company riding a wave that will crest faster than its investors expect. The $570 million is a bet on timing—whether the company can scale to unicorn-level revenue before the commoditization of AI skills makes its product obsolete. The ledger does not forget the numbers: $200 million in revenue against a $2.1 billion valuation leaves only 3–5 years to prove unit economics. If customer acquisition costs rise, or if renewal rates dip, the math collapses. For now, the cold data says: wait for the next quarterly filing before calling this a win.