Why Did Element AI Fail? US$102M Series A to a $230M Sale

Element AI failed because it raised a record US$102 million Series A before it had a product people wanted, then spent four years earning money from consulting instead of building one. Customers wouldn’t let it reuse client work as products, its costs ran far ahead of revenue, and when its 2019 Series B arrived only in part, it sold to ServiceNow for about US$230 million.

That matters to any AI founder raising in a hot market. Element AI had a Turing Award winner as a co-founder, Microsoft, Intel and NVIDIA on its cap table, and a reported US$600 million-plus valuation. By late 2020 its annualized revenue was $10 million to $12 million, according to a Deloitte opinion reported by The Globe and Mail. The buyer kept the researchers and wound down the business.

Download the data

• Element AI timeline (CSV): 22 dated events (funding, valuations, people, revenue, exit), each with its source URL.

Sources checked on 4 Nov 2026. Element AI was private. Its revenue and internal valuations come from the December 2020 sale circular as reported by The Globe and Mail, not from audited accounts. Where outlets reported different figures, each is shown with its source.

Element AI timeline

The valuation story, in five numbers

The reported figures don’t agree, and the gap is the story. All amounts are US dollars.

Before the filings, the deal price was reported three ways: TechCrunch’s sources said around $500 million, The Globe and Mail reported less than $400 million (as cited by Le Journal de Montréal), and the circular later showed $230 million. Trust the SEC filings.

Here is the worked example most write-ups skip. ServiceNow paid about $230 million for a business Deloitte valued at $45 million in enterprise value in 2020. That’s roughly 5x. On the Globe’s revenue figure, it is about 19 to 23 times annualized revenue ($230M ÷ $12M ≈ 19; $230M ÷ $10M = 23), a price you pay for a team and its intellectual property, not for a business you plan to run. ServiceNow said as much: its spokesperson told TechCrunch the focus was “technical talent and AI capabilities,” and that it expected “to wind down most of Element AI’s customers.”

Why Element AI failed: 5 reasons

1. It raised a giant round before product-market fit

Element AI announced its US$102 million Series A eight months after it was founded, calling it “the largest Series A funding round for an artificial intelligence company in history.” Co-founder Philippe Beaudoin, who led research, is direct about what that money did. “Our mistake was not seeing the tensions that could come from having too much money,” he wrote in 2022. “Money gave us the ability to bring in the best customers, but it did not pressure us into building products.”

He also defends the original bet: talent was key, and “being bold was not our mistake.” That’s a fair reading. The mistake was going a long time without the pressure that forces a startup to find something customers will pay for again and again.

2. Consulting paid the bills, so products never got the people

Element AI’s early pitch worked: big companies wanted an “AI transformation,” and the startup could bring top researchers into the room. “We had so many marquee customers,” Beaudoin writes. But “none of these products had the traction our consulting group had,” and “there was always this new customer that was too big to pass.”

The result was a split inside the company. One side wanted freedom to build products; the other wanted engineers and researchers on paying client work. Beaudoin says the tension “became really hard to manage.” The Globe, citing the circular, reports Element AI “took two years to focus on product development after initially pursuing consulting gigs.”

3. Customers wouldn’t let it turn their projects into products

This is the reason most specific to AI. A services firm becomes a product company by reusing what it builds. Element AI’s clients wouldn’t allow it. “We faced incredible resistance from the market to let us build a generic model that would then be shared and productized,” CEO Jean-François Gagné told BetaKit in September 2019. “When you get into the real money maker use case… they’re not gonna let you reuse that IP.”

Gagné also said the tooling to scale AI models, such as data annotation and monitoring, didn’t exist yet, so Element AI had to build it: “a key learning, and this is something that we haven’t solved yet.” The Globe reports the company “struggled to advance proofs-of-concept work to marketable products” and that several client partnerships faltered in 2019 and 2020.

4. Costs grew far faster than revenue

The company hired to establish what its founders called “supercredibility,” per the Globe, and its headcount grew to 500. Beaudoin lists offices in Toronto, London, Singapore and Seoul. Against that, Deloitte put annualized revenue at $10 million to $12 million by late November 2020.

The cuts came late. In May 2020, Element AI laid off 15% of staff and hired its first chief financial officer and first chief revenue officer, The Logic reported, nearly three years after its Series A.

5. The Series B came in tranches, and the second never arrived

The September 2019 round was announced as C$200 million (US$151.4 million). The circular, as reported by the Globe, shows Element AI received only the first tranche, “roughly half,” and had to meet unspecified conditions for the rest. They “were not going to be met in a timely manner.”

The alternative on offer was harsh: new money that would rank ahead of existing investors, a 12% cumulative dividend, and “other operating and governance constraints.” Management chose a sale instead. By the summer of 2020, “the company’s liquidity was diminishing as sources of capital on acceptable terms were scarce,” the circular says.

The circular also shows Element AI had asked its adviser, Allen & Co, to look for a buyer as early as December 2018, nine months before the Series B was announced.

Who got what in the sale

The payout follows a rule every founder should know: preferred shareholders are paid before common shareholders. Per the Globe’s reading of the circular:

The Globe also reports that $20 million of federal funding committed in 2020 was cancelled with the takeover.

An illustration, not the actual terms. Element AI’s preference terms aren’t public. But assume standard 1x preferences: the Series A (US$102M) plus the first Series B tranche (about US$75M) means roughly US$177M of preferred money gets paid back first. On a sale price that could fall to US$195M after adjustments, that leaves very little for common shares. A big headline price doesn’t mean founders get paid. Run your own numbers in a cap table template before you accept a valuation you may not grow into.

Lessons for AI founders

1. Raise to a milestone, not to the market. Element AI’s Series A bought talent and credibility, not proof that customers would buy a product. If a round is bigger than your next milestone needs, decide in writing what the extra money must not be spent on.
2. Put a wall between services and product. Services revenue is fine early on, but set rules: which clients you take, how many engineers can be on client work, and which engagements must feed a product. Without rules, the biggest customer always wins.
3. Negotiate data and model rights in the first contract. Element AI learned that clients won’t let you reuse models built on their core data. Agree up front what you keep (the general tooling, de-identified learnings, model weights trained on non-sensitive data) and what the customer owns.
4. Read the tranche conditions as closely as the valuation. A round paid in tranches is only as big as the milestones you can actually hit. Model the downside where the second tranche never comes.
5. Know your preference stack. Every dollar of preferred money raised sets a floor that a sale has to clear before common holders, including you and your team, see anything. The same logic applies to stacking SAFEs: see how post-money SAFEs dilute you.

For the opposite path, venture-backed growth where revenue came first, read how Cursor grew so fast. For a company that stayed profitable while it scaled, see how Canva grew so fast. More case studies live in Startup Stories.

This article summarizes public filings, a co-founder’s published account and named press reports about a private company. It isn’t investment or legal advice.