That’s how much the U.S. chipmaker shelled out to purchase the Israeli company four years ago in a megadeal that has failed thus far to generate a meaningful return on Intel’s investment.

New CEO Pat Gelsinger admitted Tuesday that his company has had a “mixed history” when it comes to acquisitions, adding Intel proved unable to convince equity markets of the rationale behind the deal.

With investors scrambling to shower money upon high-growth companies, he now believes it’s the right time to spin off what has already been an “undervalued, underappreciated” satellite flying just outside of its managerial orbit, while lifting Intel’s stock price in the process.

“As the large majority owner of Mobileye, the new operating model is frankly not all that different from what we do today,” Gelsinger told reporters on Tuesday. “We operate Mobileye quite distinctly from the rest of Intel to give it the speed and pace of operation. That will get more formalized in this structure as we look to the future.”

At the time of the deal, Intel had been steadily ceding its share of the global semiconductor market to leaner companies like Qualcomm, which focused on surging demand for smartphones rather than traditional desktop PCs. Acquiring Mobileye was seen by Intel as a unique opportunity to get in on the future of self-driving vehicles—essentially, rolling computers on wheels. 

Management at the time hailed the deal as one that would put Intel “in the driver’s seat,” creating a “foundation on which the future of autonomous driving will be built.” Now it looks as if it will settle for recouping its investment.

“They are entering a market where there’s a huge interest in any company that is developing the future of transportation,” a source close to the deal told Fortune. No longer would the Israeli company be viewed as just a simple supplier to the automotive industry, but rather a technology leader in its own right. “It’s now an entirely different landscape than when Mobileye was public.”

Late on Monday Pacific time, Intel’s board of directors, led by Gelsinger, decided to partially unwind the legacy acquisition Gelsinger inherited from the previous management team.

“That was a huge amount of money spent on Mobileye, and yet it does not look as if they ever really worked together closely,” said Stefan Bratzel, head of Germany’s Center of Automotive Management, which tracks innovation in the car industry. 

Investors cheered the news. Intel shares popped at the opening bell, gaining as much as 7.5%.

By the middle of next year, Intel plans to carve out the Israeli firm and take it public. As part of the IPO, Intel’s mobility services brand Moovit AV will be reassigned to Mobileye, effectively bundling all robotaxi-relevant operations with the new entity to make it more attractive as an autonomous-driving tech play for investors. 

The majority of the proceeds from the initial public offering of shares will flow to Intel rather than fund the new unit, but Gelsinger promised the new publicly traded Mobileye would be equipped with enough capital to ensure a “healthy balance sheet,”

The Intel CEO said he would take on the role of chairman of the board as a sign that the parent planned to retain a “clear majority” stake to profit from its acquisition in the future.

Proprietary tech

The spinoff would represent a somewhat unique opportunity for investors to buy exposure to self-driving vehicle technology suppliers. Rivals Waymo and Cruise, for example, are units within Google and General Motors, respectively.

In Tuesday’s statement, Mobileye founder and CEO Amnon Shashua praised the “accelerated growth” realized since the takeover, a period in which it nearly tripled its annual chip shipments and revenue.

It’s been clear for a while that Gelsinger didn’t entirely know what to do with Mobileye, nor how to best integrate it within Intel’s operations. The most notable cross-pollination between the firms was in the form of personnel transferring back and forth between the two companies. The technologies the two companies developed, however, remained otherwise distinct and siloed.

The writing seemed to be on the wall back in September. That’s when Gelsinger made a curious What’s-this-Mobileye-thing? kind of comment at the Munich auto show. “When I came back, it was sort of like, ‘What’s a Mobileye?’” he told the audience, before adding, “I was thrilled to learn of the extraordinary asset that Intel had added to its family.”

Analysts contend the Israeli company probably doesn’t need Intel to achieve its growth ambitions. Mobileye has long been one of the top leaders in the burgeoning field of advanced assisted and autonomous driving, supplying technology to more than 88 million cars on the road to date. According to Mobileye’s website, customers include BMW, Audi, Volkswagen, Nissan, and General Motors.

Top three

Virtually every brand has one or more models in its range that rely in some form on Mobileye’s camera sensors for assisted driving—save for Tesla. Mobileye, of its own accord, broke ties with Tesla five years ago over safety concerns following a fatal crash.

“Along with Waymo and Cruise, Mobileye is one of the top three Western players in terms of their technology, ahead of Tesla and Argo AI,” Bratzel said, adding China also has three strong competitors in Pony.AI, Baidu, and AutoX. 

Mobileye relies on its own proprietary technology rather than on incorporating know-how from Intel. Eight units of its deep-learning chipset, the EyeQ5H, will power all the sensing, mapping, and path-planning for its upcoming robotaxi service in Germany. 

Shashua told reporters in September he aims to deploy a fleet of about 50 vehicles in Munich that will forgo the use of a safety driver by the end of 2022. Mobileye will then be among the first to take advantage of the German government’s recent decision to legalize autonomous cars across the country.

Fivefold growth

By comparison, in the U.S., the rollout of the technology is impeded by a lack of federal legislation, with each state responsible for setting up its own laws governing the use of fully autonomous self-driving cars.

Alphabet unit Waymo has been operating a pilot program in Phoenix, while Cruise, controlled by General Motors, is attempting to launch a service in San Francisco under far more challenging traffic conditions. Not to be left out, Ford and Volkswagen-backed Argo AI hopes to launch a pilot in Miami soon.

Tesla, by comparison, has focused on deploying a fleet of cars that it then upgrades over time to enable full self-driving. This is a controversial approach, however, not least because of its reliance on U.S. customers to develop the technology rather than professional safety drivers.

Spinning off Mobileye doesn’t mean that Intel has no ambitions to serve the automotive market. Gelsinger anticipates the amount of chips built into an upscale passenger car will quintuple by 2030, incorporating a fifth of the overall bill of materials (BOM). He estimates the total addressable market for automotive semiconductors will grow to $115 billion by the end of this decade, up from an estimated $50 billion in 2021.

Courtesy of Intel

Both of Intel’s two main U.S. rivals are likewise keen to expand their footprint in this burgeoning market. Nvidia counts Mercedes-Benz and Volvo Cars as customers of its autonomous driving chipsets, while Qualcomm recently inked a deal with BMW in addition to its leading position in semiconductors that run automotive infotainment features.

Intel, however, hopes to benefit from a key differentiator: It, like Nvidia and Qualcomm, designs processors, but unlike the duo it also manufactures them. Gelsinger believes this uniquely positions Intel to help mitigate the current crisis in automotive chip production, given its two so-called fab-less rivals do not have manufacturing capacity of their own.

As a result, Gelsinger came to the Munich auto show in September not only to celebrate the announcement of Mobileye’s robotaxi plans, but to pitch his Intel Foundry Services Accelerator program that migrates carmakers to considerably more advanced chips than those found in today’s cars.

“The role of the semiconductor supply chain is dramatically increasing, not only today but, we expect, for the next decade and beyond,” Gelsinger told reporters.

This story was updated to include comments made by Intel CEO Pat Gelsinger at a Tuesday afternoon news briefing.