Atomic's SEC filing shows $18.5 million sold against a $12.5 million Series A

The $6 million gap between the announcement and the filing ^[1] has at least three plausible readings. The round may have grown after the release was drafted, with money from investors beyond the two leads. Early money from DVx Ventures, the firm founded by former Tesla President Jon McNeill where Atomic was incubated ^[11], could have converted into the same sale and been counted there. Or the filing may bundle more than one close. Whichever it is, the filed figure is 48% larger than the announced one ^[2]. Put the other way round, a runway worked out from the press release uses about two-thirds of the cash on record ^[3].

Good Chop's result is the one with a cash consequence. Cutting inventory cover to four weeks ^[6] takes stock turns, on a 52-week year, from roughly six a year to 13 ^[4]. Each dollar of inventory at the HelloFresh unit now supports more than twice the sales it did ^[4].

Every customer figure in the announcement comes from Atomic, and neither DashMart nor Good Chop is quoted ^[7]. The praise on the record comes from a lead investor. "We've never heard customers talk about a software vendor the way they talk about Atomic. They trust it with daily operating decisions and are asking for more," said Daniel Klass, founder and managing partner of Klass Capital ^[10].

Atomic is also not asking customers to replace their enterprise resource planning software. Its planning layer runs alongside it ^[8]. That keeps the sale small and leaves the system of record with another vendor. Michael Rossiter, the co-founder and chief executive, pitches it to companies that never built what his team built at Tesla ^[3]. "They're still running critical parts of their business through legacy software, hundreds of spreadsheets and individual heroics," he said ^[9].

Against rivals the check is modest. Freehand raised $75 million and Lyric $43.5 million for AI supply chain products ^[12]. Even the filed $18.5 million is under half of Lyric's round and about a quarter of Freehand's ^[5]. I think investors paid for a working reference account: one large customer where the software already places most orders. In an enterprise sales cycle, that counts for more than a bigger fund. The counter-case is concentration, because the customer supplying most of the evidence could take it away by leaving. The company's account does not say whether the 90% ^[4] is counted in orders or in dollars of spend. If it turns out to be a count of routine reorders covering a small slice of DashMart's spend, the reference account is weaker than the round assumes.