Andela's original model was a genuinely proprietary one: train junior engineers in African hubs and rent them out. Training is a fixed cost recovered over the length of an engagement, which means the model only works if buyers keep wanting juniors. When COVID-era demand moved to seniors, that assumption broke, and the layoffs followed directly — ~400 developers cut in 2019, 135 staff (10%) in 2020, with further cuts in January 2025 (Grokipedia).
The pivot that saved the company deleted the reason to own it. Moving from train-and-place to a global open marketplace removed the only asset Andela had that a competitor could not buy — trained supply — and left a commodity staffing spread in a market where the public comps trade at 0.2–0.8x EV/Revenue (Multiples.vc). See What the public market pays for labour and What the model actually is.
The three models, in order
- 2014–2019 — bootcamp plus placement, African hubs. Recruit, train, embed. Proprietary supply, high fixed cost, junior-weighted.
- 2020 — fully remote, offices closed. The physical hub was the training apparatus; closing it ended the differentiated supply.
- 2021 onward — a global marketplace across 135 countries. Open supply, no training moat, competing on price against every other staffing platform.
Peak mark was $1.5B on a $200M Series E led by SoftBank Vision Fund 2 in September 2021 (Sacra) — set at the top of the remote-work bubble, one pivot after the model that justified a premium had been retired.
What it charges
Andela bills $20–40/hour for junior engineers and $50–100/hour for senior engineers (Grokipedia). That is a staffing spread, not a take rate: the buyer pays a bill rate, the engineer is paid less, and the difference is the business. Any "revenue" figure for this model is bill rate, and a multiple computed on it is computed on the wrong number — see GMV is not revenue.
For comparison, Toptal bills $60–150+/hour blended on the same shape of transaction (TheFrontendCompany) [WEAK], and raised $1.4M rather than hundreds of millions.
Recent activity
- Carrol Chang (ex-Uber) became CEO in September 2024 (Forbes AU).
- Acquired Woven in January 2026 for AI-based engineer assessment (Grokipedia) — an attempt to rebuild a vetting asset by purchase after having dismantled one by pivot.
- Toptal sued Andela in May 2024 over trade secrets and the alleged poaching of 30+ staff; the case was still ongoing as of November 2025.
What it cost
The honest accounting is not the layoffs. It is that a company which raised $330M+ and carried a $1.5B mark has disclosed no revenue in five years, and the only public events since 2022 are a CEO change, a small acquisition and litigation. That is the shape of a business that is alive and is not compounding.
The generalisable lesson is about what kind of scarcity you own. Andela's training pipeline was the sort of supply that money alone could not assemble quickly — the highest-value thing in this atlas's framework. Marketplace supply, by contrast, is supply anyone can convene with a budget and a quarter. Trading the first for the second bought reach and cost the moat, and the valuation was set before the trade was visible in the numbers. See Which side you build first and Where the supply can legally live.
The sources disagree on how much Andela raised. One puts it at ~$330M (Sacra); another at ~$381M across 8 rounds (Grokipedia). The $200M Series E and the $1.5B mark are consistent across both. The total is unresolved here.
The two records also disagree on whether anything has happened since 2022: one describes the company as having produced no funding, revenue or major coverage 2023–2026, while the other documents a September 2024 CEO appointment, January 2025 layoffs and a January 2026 acquisition. The second is more specific and more recent, and is the version used above.
Revenue, headcount, gross margin, engineer pay rates and customer concentration are all undisclosed and were not found. The layoff figures come from a single aggregated source. Distinguishing "quiet" from "dying" here needs primary work — company filings and headcount trends — that this research did not do.