Slashdot Mirror
Uber Loses At Least $1.2 Billion In First Half of 2016 (bloomberg.com)
An anonymous reader writes: The ride-hailing giant Uber Technologies Inc. is not a public company, but every three months, dozens of shareholders get on a conference call to hear the latest details on its business performance from its head of finance, Gautam Gupta. On Friday, Gupta told investors that Uber's losses mounted in the second quarter. Even in the U.S., where Uber had turned a profit during its first quarter, the company was once again losing money. In the first quarter of this year, Uber lost about $520 million before interest, taxes, depreciation and amortization, according to people familiar with the matter. In the second quarter the losses significantly exceeded $750 million, including a roughly $100 million shortfall in the U.S., those people said. That means Uber's losses in the first half of 2016 totalled at least $1.27 billion. "It's hardly rare for companies to lose large sums of money as they try to build significant markets and battle for market share," said Joe Grundfest, professor of law and business at Stanford. "The interesting challenge is for them to turn the corner to become profitable, cash-flow-positive entities."
"Amazon's initial business plan was unusual; it did not expect to make a profit for four to five years. This "slow" growth caused stockholders to complain about the company not reaching profitability fast enough to justify investing in, or to even survive in the long-term. ... It finally turned its first profit in the fourth quarter of 2001: $5 million"
Seven years it took to hit profit, but they knew that and said it would be like that all along.
I suspect that Amazon's turnover and revenue were significantly higher than anything Uber's ever seen, and I suspect they never lost $1.2bn at any stage of their inception.
It was also - as stated - highly unusual.
If their goal is to collect data for autonomous car research they're doing a piss poor job of it. "Comma.AI" has an open source dataset published and I KNOW it didn't cost them $1 billion dollars to make.
The future of autonomous vehicles is doing more training in video games.
Right now, training a basic convolution neural network on ImageNet requires 8-12GB of GPU memory. That's a $1200 investment to get your feet wet!
256x256x3 tensors are NOTHING compared to the 1080p-4K images which are trivial to collect with today's technology. Until the performance of convnets trained on 256x256 resolution images improves: there's not much point in scaling up the networks.
Last MLP I ran on a much smaller image took 33 seconds to process on a GTX 970. Those frame-rates are not conducive to reasonable servo controller's at 60-90 mph. This is no-doubt why the original Google vehicles had such low top-speeds.
There's a relationship between algorithms and hardware and we need much better algorithms, much better hardware, or some combination to make autonomous vehicles more than just science experiments. Just my humble opinion from the trenches...
It was a lot more than that. The figures vary but this article puts the loss on the original Xbox at $5-7billion:
http://n4g.com/news/1789459/former-xbox-head-says-original-xbox-lost-between-5-billion-and-7-billion
They then launched the 360 and immediately had to allocate $1billion to fixing units that had RROD.
EBITDA is commonly used as a proxy for operating profit in the VC community.
Depreciation & amortization aren't going to be hugely relevant for Uber (unlike the companies Buffett tends to invest in), and in the context of startups where capital expenditures relate to growth not replacement, and are going to be sourced from further investment rather than retained earnings, these noncash expenses just clutter the laser focus on operating profits, which are the overriding interest... the first thing you want to know is "when are they going to stop burning cash", not "how long will it take them to recoup the investment on their IT infrastructure"
Interest expense gets taken out because it reflects corporate finance decisions, not operating results. The business isn't doing better or worse because a company relied more or less on debt financing vs. equity, or had to pay a higher or lower interest rate. It's not like anybody's ignoring the debt, it's just that people whose business is to change the capital structure prefer to value the entire enterprise in a capital structure agnostic manner, then deduct the full value of debt to get what's left over for equity... a balance sheet approach, rather than GAAP's approach of anointing net earnings on the income statement as the One True Number that we need to slap a multiple on.
EBITDA is not a finishing point in evaluating & valuing companies, it's just the point along the way that's of the keenest interest.