When facing bad public relations, one of the oldest plays in the playbook is to try and change the narrative. Execs are departing? You’re being sued over possibly stolen technology? Stories of sexual harassment in the workplace are making national news? How about you openly share some of your heretofore mostly secret financial performance to get people focused on some cold, hard numbers, instead? Whether this is literally what went on in the minds of the executive team at Uber, yesterday’s data dump showing that Uber booked a staggering $20 billion in rides last year certainly got a lot of attention.

Unfortunately for Uber, the financials shared with Bloomberg, paint a decidedly mixed picture beneath that eye-popping figure. While Uber’s growth has been remarkable and remains very strong as it approaches 7 years old — business more than doubled in 2016 — the company has yet to demonstrate a sustainable or profitable business model. Perhaps worse though is that despite at least $8 billion in cash burn Uber has little in the way of a strong competitive moat to show for its efforts.

First, the good news

In 2016, Uber managed to record revenue of $6.5B on those $20B in bookings. That’s a “take rate” of more than 32%. Over time, Uber has been able to record a higher percentage of the cost of each ride as revenue, typically giving drivers a smaller slice of the fare. Drivers used to typically receive 80% of each ride’s cost but now are clearly averaging a lower figure.

It’s also true that growth remained strong throughout last year. Uber told Bloomberg $6.9 billion of the bookings came in Q4 and from other data provided made it possible to impute $5.4 billion was booked in Q3. That means just $7.7 billion was from the first half of the year and the final three months were nearly equal to the first six. That kind of growth isn’t just healthy it’s meteoric. By way of comparison, Facebook added about $10B to its top line last year, but that was from $17B –> $27B. Uber managed more absolute dollar growth on a smaller base. It’s likely there’s much more growth to come.

Indeed these booking figures have been soaring for years, from about $3B in 2014 to $9B in 2015 to last year’s $20B. NYU finance professor Aswath Damodaran, a longtime Uber skeptic, suggested back in 2014 the entire global rides-for-hire market was $100B. While the growth of Uber has doubtless hurt incumbents like taxi services, it’s actually unlikely it has taken a 20% global share. Instead the combination of innovations Uber has delivered — on-demand rides through an app with route tracking, upfront pricing, et al. — has allowed it to grow the market in for-hire rides.

The figures above are all, incidentally, exclusive of China which Uber left last year through a sale of its business to rideshare giant Didi. Uber had been burning $1 billion+ in China for the past two years. The drain of that business is now a footnote on its financials and, perhaps equally importantly, allows the company to concentrate management attention elsewhere. While exiting China might not have been the desired course of action, it doubtless leaves Uber better to compete in India, Latin America, and Europe.

Next, the not-so-good news

All those gains in bookings are fairly easily to understand but breaking it down into financial performance is much harder. For accounting reasons, Uber correctly only records its “take” as revenue under Generally Accepted Accounting Principles, or GAAP as it’s known. At least it does this most of the time. The accounting fiction Uber has operated under is that the real transaction is between rider and driver, with Uber merely a market maker. On traditional Uber rides that means the piece it claims is a “license fee” for use of the software and underlying marketplace. Priceline and Groupon similarly only book their slice of the pie as revenue, not the total amount of the hotel room or discount voucher they sell.

What’s concerning with Uber is that something very different happens with UberPool, the company’s shared-car service. It’s an essential component of growth in that it allows for more riders without an increase in drivers. But on UberPool, the company told Bloomberg, “the company counts the entire amount of an UberPool fare as revenue.”

I’ve sent a request into the company asking to clarify why, but based on what I do know here’s the likely reason: With UberPool, the company often charges a flat rate irrespective of distance traveled or otherwise discounts the offering. It then pays the driver based on a formula accounting for time and distance whether or not the ride ends up being shared. (For those unfamiliar, if you request an UberPool and you’re the first person picked up, there’s a chance you will never be matched with another ride and end up riding alone. You don’t pay more when this happens, but Uber is on the hook for standard UberPool rates to the driver.)

In these cases, Uber is basically paying the driver a rate for work and charging the rider something often very different — there’s no way to claim this is any sort of license fee. The gross margin on UberPool is likely quite low, then, given that rides can be as low priced at $3-5 in some markets. But taking the price paid as revenue necessarily inflates that figure versus a traditional Uber ride even if said ride costs 2x as much. Example: Rider pays $5 for an UberPool ride, revenue = $5 — that’s often $5 x 2 or $10. By contrast, rider pays $10 for an UberX ride, Uber’s revenue averages about $3.20 but the margin is only impact by the costs of payment processing, fraud, insurance, etc.

With all that, Uber grew its reported revenues from about $1.8B to $2.9B over those last two quarters of 2016. It was able to translate more than 70% of bookings growth into actual revenue and saw its loss only expand by about $60 million. But what a loss it was; Uber bled $991 million in red ink just in Q4. And here’s where that gets scary: Of $2.8B in losses last year, more than 1/3 were incurred just in the fourth quarter. What that demonstrates is though the ratio of losses : revenue may be shrinking, the absolute size of the loss continues to grow.

With a self-reported $7B+ in cash still on hand and another $2.3B in available credit, Uber isn’t in any short-term danger of a cash crunch. But the fact that losses have still grown with revenue suggests something approaching $1B each quarter is a possible scenario for 2017. There is no precedent for these kinds of losses in startup history. Worse still is that seven years into the Uber story, it’s reasonable to argue whether this business model works as currently constituted.

On Twitter, startup investor and AngelList founder Naval Ravikant suggest this isn’t a problem: “If Uber raised prices 15%, it’d be profitable.” Except it isn’t that simple. If Uber raised prices 15% and kept 100% of the increment, it would’ve earned an additional $3B last year — perhaps just enough to offset its losses. If Uber raised prices and — in a more likely scenario — only kept the 1/3 it normally receives, the extra $750M would have covered about 1/4 of the company’s loss.

Then, of course, there’s the issue of whether a price increase so large would have no impact on demand. Everything I know about the company says instead that it would. Uber routinely has used lower prices to expand the number and frequency of riders. Few goods are truly inelastic and on-demand rides don’t fit the model well. When they are inexpensive enough, people will find themselves in Ubers often. But if prices rise, options like transit, driving one’s own car, biking, et al. begin to look much more attractive. Even if volume fell just a few percent with such a large price increase, the gap that would leave would be in the nine figures.

And finally, the worst news

Underlying the assumption that Uber even could raise prices is the belief that it has created a competitive “moat” that gives it a path toward monopolistic power. There is scant evidence that’s true, however. Uber has strong competition around the globe, from Lyft in the U.S. to Ola in India and a major player working against it nearly everywhere else. In attractive markets like New York City, there are numerous options, including Gett, Juno, and Via.

Uber fans tend to draw comparisons to Amazon, which long made a habit of money-losing quarters and has seemingly emerged on the other side as an unassailable force, slowly crushing its retail competition. But Amazon spent much of the last two decades creating its fortress, which consists of things like a network of distribution centers located ever closer to customers, more power over suppliers who have little choice but to sell on Amazon, and arguably the greatest loyalty program ever created in Amazon Prime.

For all the success of Uber, it has few assets to show for those billions in losses. Beloved by many customers, its brand is also among the more hated in technology. Anecdotally, I know dozens of people who will simply never do business with Uber. And the recent firestorm surrounding the company over that began with the Susan Fowler story and escalated over President Donald Trump and his travel fan has doubtless added to the #neverUber rolls. The exodus of top-level executives is rarely a good sign for recruitment; and at Uber it’s a lengthening list.

Further, while it does have an excellent technology suite, there are still too many times when Uber offers a frustratingly inaccurate estimate of wait times and even untrustworthy pricing on some occasions. The closest thing to Amazon’s assets is that Uber has the longest roster of drivers in nearly all markets. That bigger supply leads to shorter waits and greater availability than competitors.

Unfortunately for Uber, it’s often at odds with those drivers as it cuts rates, pushes UberPool (where drivers are convinced they earn less), and fights them in court over whether they ought to be treated as employees. Drivers very frequently take advantage of their freedom as independent contractors to work for multiple services and few express much loyalty toward Uber. Further, unlike an Amazon distribution center in Northern Virginia that might serve the D.C. metro area for decades to come, Uber’s driver pool consists of a lot of transient part timers, many recent immigrants who will likely be in shorter supply for the next several years, and retirees who typically won’t be Ubering forever.

Yet even if Uber somehow turns around its driver relations, a goal of the recently departed executive Jeff Jones that seems far distant, it still faces a reality that other recent tech giants haven’t deal with: Uber is not a returns-to-scale business. Facebook’s giant user base allows it to sell more and more advertising without significantly growing its headcount. Netflix might spend millions on a new show but can stream it to new users for pennies without paying any extra for the content. And, Amazon, for all its reliance on old-school logistics like UPS, physical goods management, and the like can ship more packages to more people over time. Partly that’s robotics, partly it’s more efficient methods, partly it’s through acting as a market and taking a cut of others’ sales. It also can fund its retail ambitious through Amazon Web Services, it’s dominant hosting provider.

Uber looks more like a ground-based airline: It can’t carry more people without more drivers (planes). Once it fills all the seats in an UberPool (maximizes yield), it has maxed out returns. It can’t raise prices with viable competitors around (unless they match the increases). And the market for new drivers is constrained (perhaps not quite as severely as aircraft or gates at airports) by population growth, availability of other attractive jobs, affordability of vehicles, etc. Uber can do little or nothing to mitigate those.

Where does this leave the company? In a rather uncertain place. Those betting on self-driving cars to solve Uber’s economics down the road are assuming that competition there won’t be robust, which seems far-fetched. And in the meantime, the company is literally losing billions every year without demonstrating that its business should continue to receive the kind of investor faith it has to date. While this doesn’t mean Uber is going away, it also casts doubt on whether it’s poised to deliver on its $69B valuation anytime soon.