Google’s famed corporate research lab, first launched as Google X, was a sandbox for the company founders and its band of experimental scientists. When Google became Alphabet, the lab announced a sparer name and a leaner mission: To transform those sandbox projects into real commercial operations.

On paper, that is happening. News came out on Friday that the unit had hired a seasoned online travel exec for its self-driving car team. A few days earlier, it recruited a broadband veteran to head its Project Loon internet balloon initiative.

But the reality is murkier.

Several people who have recently left X and those close to it describe the Alphabet unit as sputtering, unable to bring projects to life. They say the issues at X aren’t technical hurdles, but a combination of red tape and knotty internal politics. Rather than accelerating the “moonshot factory,” sources say the Google-to-Alphabet reshuffle has clogged it up.

The Alphabet company that is meant to create new Alphabet companies is struggling to get them out the door. And I have heard repeated talk of a coming shake-up at the division.

A rep for X declined to comment.

In the past, Alphabet had budgeted more than $1 billion a year for X, according to one high-level source*. Most funds went to its car project, now seven years old, and Google Glass, the much-hyped wearable that fizzled in its first incarnation and left X last year.

But hiring slowed to a crawl when the Alphabet reorganization arrived, sources said, part of a drive to evaluate spending on the company’s ambitious offshoots.

And people started leaving. The most high-profile exit was a cadre of self-driving car engineers who formed the startup Otto earlier this year, then sold it to Uber this month.

Multiple people who have left X told me that the inability to ship products was a leading reason for the departures.

Multiple people have also told me that Astro Teller, the longtime X chief, is increasingly frustrated. Sources describe most of X’s public projects — Project Loon, drones, robotics and wind energy kites — as rudderless.

X still employs many gifted roboticists, designers and engineers. Google founders Larry Page and Sergey Brin are both actively involved with the teams, sources said. They moved their Alphabet offices to the third floor of the X building, according to one source.

“It’s Sergey’s Batcave,” said one former employee.

The Batcave has produced spinouts that stuck around, including the Google Brain research unit and its Verily life sciences arm.

Many other projects won’t see daylight, but that’s partly by design. Racking up failed projects is the modus operandi of X. Before the Alphabet reorg, Teller implemented a bonus system rewarding employees for nixing their own projects, designed to funnel resources to feasible efforts.

But people who have left said projects are more likely to die because of organizational inertia — not because their leaders kill them off themselves. They also complain about constant changes in direction from different project leaders. Mike Cassidy, who stepped down from Loon, ran the team “like a fire drill,” a former employee said.

Some frustrations coming from X echo those across the Alphabet companies, such as connected device maker Nest.

There’s no clean, simple playbook for trying to birth entirely new things, especially when extremely intelligent and passionate people are involved. (See: Google’s unwieldy robotics team or any messy startup story.)

“The moonshot factory is a messy place,” Teller has written and said multiple times.

And while the car project under Teller has lost several critical members and faces criticism for dawdling, it is expected to “graduate,” X’s parlance, soon.

X is still tinkering with moonshots. I reported that the unit phased out one virtual reality headset, but others have reported, and I can confirm, that another VR thing is in development.

I wish I could tell you more. But X remains opaque — to outsiders and insiders. Rank-and-file Googlers cannot even use their company badges to access the X offices.

Besides, the entire lab more or less shuts this coming week. Virtually all the execs go to Burning Man.

* Operating losses for the Alphabet parts outside of Google topped $3.5 billion last year.

Anyone with insight into X can message me securely on Signal, Telegram or WhatsApp at 917-917-0921.

This article originally appeared on Recode.net.

For the past year, Ruth Porat, the CFO of Google and its parent Alphabet, has told Wall Street that Google Fiber is her most expensive unit outside of the core business — and is well worth the costs.

Her bosses may be telling Fiber employees the opposite.

According to a report in The Information, Alphabet chiefs Larry Page and Sergey Brin recently instructed Fiber to severely trim staff and expenses, frustrated with mounting costs of delivering high-speed internet by digging up dirt.

Creating broadband networks via traditional pipes is enormously expensive. And Fiber still hasn’t proven that it has figured out a better way to do it. The Information story comes on the heels of reports that Fiber has put plans to build broadband networks on hold in two cities as it ponders ways to roll out experimental wireless tech.

A rep for Fiber declined to comment.

Fiber, like the “Other Bets” businesses outside of Google, is facing ongoing scrutiny about its operations. Here are the key parts of The Information’s report.

• The unit initially shot for five million broadband subscribers in its first years, but has fallen short of that.
• Last month, Page told Craig Barratt, the CEO of Fiber (or Access, as it’s known), to halve his staff down to 500.
• Page also told Barratt he wanted to cut the cost of delivering internet to homes down by a tenth. Industry experts had told me that wireless plans, like the one Fiber is pursuing, could be about a fifth as cheap as fiber networks. (That said, Page is renowned for his near-impossible demands.)
• Porat, who has developed a reputation as a cost cutter, interceded on Fiber’s behalf, arguing to Page that Fiber’s business model is defensible.
• Barratt considered leaving earlier in the year, reportedly irked by the changes at Alphabet. If he did, he would not be the first “Other Bets” exec to do so.

We reported earlier that Fiber’s current expansion model costs at least $1 billion per market.

While Fiber has hit pause in two markets, Portland and San Jose, it is still moving forward with broadband and cable connections in others. Earlier this week, it announced that service was active in Salt Lake City, its seventh metro area.

This article originally appeared on Recode.net.

There’s another executive mix in the Alphabet soup.

Tom Moore, a senior VP at the broadband satellite company ViaSat, is taking the reins at Project Loon, the ambitious effort to string high-altitude balloons into a global internet network.

Mike Cassidy, a serial entrepreneur who has helmed the project for the past two years, is stepping down from the position but remaining at X, the company under Google parent Alphabet that houses the project.

Recode had previously heard that Cassidy was stepping away from the project, which has yet to deliver a commercial product. But we didn’t have all the details.

The hiring of Moore, a communications industry veteran, signals that the project may become more reliant on satellite networks. Moore’s company, WildBlue Communication, which he founded in 1998, was acquired by ViaSat in 2009.

It’s also a similar pattern to other parts of Google’s far-fetched research lab. The Alphabet arm has hired leaders with commercial experience as they try to turn their projects from science experiments to viable operations. In the fall, they hired a former Hyundai exec to run its self-driving car unit.

Still, the Alphabet divisions have lost some key executives over the past year.

Astro Teller, the head of X, had this to say in a statement about the new hire: “Under Mike’s scrappy, entrepreneurial leadership, Loon moved from science project to viable venture, and Tom’s valuable industry experience will help launch us into the commercial stage of this moonshot.”

In the fall, Loon announced it was starting testing with carriers in Indonesia, its fourth country partner, despite technical hurdles that nearly nixed the project.

Cassidy said in March that Loon was “getting close” to commercial deployment. Cassidy took over the project in the fall of 2014, after spending three years as a director at the Google X research lab.

He arrived at Google after an entrepreneurial streak, founding and selling three startups. His fourth, an online travel guide called Ruba, became a part of Google in 2010.

Moore will start in mid-September as a VP at X, the company said.

This article originally appeared on Recode.net.