For all its heat, the fire that is Uber in 2017 hasn't scorched everything. While the nigh-apocalyptic past six months have felled founder and CEO Travis Kalanick, and sparked questions about its ability to keep its employees safe, let alone happy, Uber's self-driving car program seems to be doing just fine.

It's a rare but vital bit of good news for Uber, for which autonomy is an existential question. If another company figures out how to operate a taxi service without paying drivers and Uber cannot, it's lights out, unicorn. “What would happen if we weren’t a part of that future? If we weren’t part of the autonomy thing?” Kalanick told Business Insider last year. “Then the future passes us by.”

Now, Uber's self-driving program hasn't been unscathed. It faces a vicious lawsuit from Waymo, Google's self-driving car spinoff, which accuses it of using stolen IP to advance its autonomy research. Last month, Uber fired Anthony Levandowski, its self-driving car lead who allegedly brought that IP over from Google. In the past year, it has lost talented employees to competitors like Argo AI, owned partly by Ford, and Aurora, a startup run by former Google autonomous vehicle head Chris Urmson.

But researchers, roboticists, and entrepreneurs in the small automation community say Uber remains stuffed with smart, hardworking people, and that the research is going forward at Uber's Advanced Technologies Center in Pittsburgh, Pennsylvania. There, 400 employees race to deliver the autonomous future, branded with a big "U."

“I can say without reservation that Uber built something really special in the lab there,” says David LaRose, who spent two years at Uber ATG before leaving earlier this year to become chief scientist at the manufacturing robotics company Carnegie Robotics. “It’s an amazing place with a lot of great people.” (Uber declined to comment for this story.)

And engineers are still hyped about getting self-driving cars on the road. “You do feel it around town,” says LaRose. “There's a snap in the air that feels like something big is happening.”

Uber landed in Pittsburgh in 2015, when it announced its intention to become a self-driving force. It poached more than 40 researchers from Carnegie Mellon University, which has one of the country's best robotics programs and a long history in autonomy.

Suddenly, scientists who had toiled for decades on projects many considered irrelevant had greater opportunity to bring their technology into the real world, where it could generate profits and save lives. (It helped that the Steel City's lower real estate prices and booming arts and restaurant scene made it a real nice place to do cutting-edge robotics work.) In less than two years, evidence of their efforts cruised about town, in the form of self-driving Ford Fusions.

“When Uber first came into town, there was no one doing what they were doing,” says Jackie Erickson, who founded the Pittsburgh Robotics Network last fall. “They could pick the brightest people to work on the something that was extremely meaningful, the hardest problem in the world.”

The rosy picture has faded a bit. In the past year, amid reports of clashes between the self-driving team in Pittsburgh and the trucking-focused unit run out of San Francisco (formerly helmed by Levandowski), some of that top roboticist talent left. Peter Rander, one of those poached CMU roboticists who became engineering lead at the ATG, founded Argo AI, joined by autonomous vehicle engineer Brett Browning. Aurora picked up Drew Bagnell, Uber’s head of perception and computer vision. But turnover is typical in booming industries with limited talent pools. “The competition for talent is fierce, at all levels, but there are companies who are really interested in getting senior people who have been in the field for 15 years,” says Steve DiAntonio, president and CEO of Carnegie Robotics.

And plenty of smart people remain at Uber. John Bares, the former head of the National Robotics Engineers Center, is still Uber ATG operations director. Uber hired AI superstar Raquel Urtasun (and some of her researchers) out of the University of Toronto last month, launching another R&D outpost in the Great White North.

Meanwhile, local robotics researchers and entrepreneurs say Uber’s crises don't get the same kind of play in Pittsburgh they do in Silicon Valley. “It hasn’t been sensationalized here,” says Jorgen Pedersen, CEO of the robotics company RE2 Robotics. “People are just heads down, engineering, solving hard problems, and making progress.”

Researchers bound to Uber have good reasons to stay put. First, autonomous vehicles take years to develop. “If we were talking about a technology that anyone could just whip up in their basement over the course of a couple of weeks, maybe people would be making more moves,” says Mike Wagner, Pittsburgh-based co-founder and CEO of Edge Case Research, an automation software testing and consulting company.

Second, building a safe autonomous car requires vast datasets, the kind that take manpower to build, and fleets of vehicles to test your tech on. You also need a teams, where each member is responsible for handling a different self-driving challenge—computer vision, the mapping, hardware, artificial intelligence. For now, Uber is one of the few companies that can provide the cash to fund all that, and keep it flowing.

That longer time horizon also makes engineers less likely to jump ship based on moments of rockiness and instability. “Engineers evalsuate the likelihood of success at a particular company over the long run and they may be less moved by the crisis of the day,” says David Kalson, a lawyer at the firm Cohen & Grigsby, who advises robotics companies in the Pittsburgh area.

Plus, the Silicon Valley glorification of the disruptor doesn't play everywhere. “It’s not just about the entrepreneurial instinct," says Wagner. "It’s really about the technology. Engineers see where they can make the most continuing impact." Abandoning the company giving you a chance to do that is a little more challenging and a little less appealing in that light.

Yet Uber’s self-driving program has many challenges ahead. Employees tire of the constant shuffling up top. Workloads can be onerous. Fatter paychecks might be out there. And the partners Uber needs to bring a product to market may not want to work with the scandal-covered company.

Then, of course, there’s Waymo, and the lawsuit that got Anthony Levandowski fired in the first place. Uber has dodged the worst of it thus far, avoiding a preliminary injunction that could have paused or even shut down its self-driving program. But if Waymo wins this one, it threatens to cut off something that autonomous vehicle research needs desperately: the money. No checks mean no engineers means no self-driving cars. No self-driving cars means no future Uber. For now, at least, the self-driving cars roll along.