That sounds taken out of context, or misquoted. Its reasonable to talk about the marginal profitability of an activity. E.g. Given employee base (shopper workers already hired), current app and backend (no marginal cost for developers/executives) then the sales price minus cost-of-sale was positive. Very important number! Means the company would be profitable after scaling that part of the business enough to cover fixed costs.
Most startups are actually looking for that magic formula. They spend spend spend until they find it, then scale scale scale to become a profitable business as a whole.
Doesn't look like a misquote to me. And while, yes, unit economics are a thing, there can be a huge gap between being unit profitable and profitable as an organization. A bunch of delivery companies flew into the ground during the first crash under the same circumstances (including a few grocery delivery companies).
If you're running a company with 500 employees an a big office in San Francisco (where employees average ~$100k a year, fully loaded), and each of your deliveries nets 1% of a $50 order, on average ($0.50; not a ridiculously low net margin for the grocery industry, even in logistically optimal scenarios -- which delivery is not), you've gotta be doing (500 * $100,000) / .5 = 100 million sales a year just to break even. AKA, $5 billion a year revenue run rate.
So then you say: "OK, we'll just cut some of those expensive SF people, and we'll bring the curves closer together!" And that could happen. Or you could discover that getting those margins was only possible with X million sales a year, and getting those requires at least 500 employees to run operations without dropping the ball. And then your investors stop throwing money at you, because the business economics look scary, and the funding climate has changed. And then you die.
Again, this is not a made-up story.
When huge investors get involved in land-grab businesses before they're profitable, they're all betting that their horse will be the next Amazon. But there's only one Amazon. And even Amazon isn't that profitable. And Amazon started by competing in a high-margin industry.
While I agree that marginal probability is important, I think it is a fallacy to assume all of those other costs don't have some variable component as well. There will always be shopper worker churn that will grow as the business grows, additional (albeit low) costs in software to scale and add new regions, etc.
In a business with sizable margins, it may be OK to discount some of these other things, but in a business with teeny tiny razor thin margins like grocery delivery, one should have a very healthy skepticism about hand-waving away real costs.
Most startups are actually looking for that magic formula. They spend spend spend until they find it, then scale scale scale to become a profitable business as a whole.