• Could consolidate 50+ ECUs typical in new EVs to just a few
• Nets a 3-5% gain in efficiency and range, faster charging
• Simplifies assembly, cuts weight, extends reach of OTA updates
• Karma Kaveya EV arriving in 2026 is built on Intel architecture

The idea of the software-defined vehicle is, it seems, a billion-dollar buzzword in today’s auto industry. 

Truly, the money’s on it. In November, Volkswagen—recognizing Rivian’s progress in the area—invested up to $5.8 billion in a new joint venture with the U.S. EV maker. The investment is aimed at creating both the electric architecture and the software for what might amount to millions of next-generation software-defined vehicles (SDVs).

While we wait and see what that partnership brings, there’s a widespread consensus among tech suppliers and automakers that SDVs are the future for EVs and beyond—while few have provided a concrete roadmap of how and when they intend to get there.

At CES in Las Vegas earlier this month, Intel stood out as an exception, for its unveiling of a comprehensive, whole-vehicle SDV solution. It claims much more than over-the-air updates, and that it might make EVs not only more adaptive to needs but also more efficient—with up to a 5% gain in EV range and efficiency from its new software-defined layout plus a more responsive driving experience. 

Karma to be “living development prototype” for Intel architecture

That layout isn’t a far-out concept, and it could be put to the test in less than two years. The Intel whole-vehicle software-defined architecture is due to be incorporated on the Karma Kaveya, a 1,000-hp fully electric all-wheel-drive coupe, with claimed 0-60 mph acceleration under 3.0 seconds, that’s slated for production in 2026. 

“We think that we have the opportunity to be the very first manufacturer to deploy a full ground-up true software-defined vehicle architecture,” said Marques McCammon, at Intel Automotive’s CES presentation. “And we’ll do that with the Karma Kaveya; it will have the ability to have workloads redistributed, be updated via the cloud, and it will serve as a platform for every Karma vehicle going forward.”

McCammon said that the Karma EV will be “a living development prototype for the broader industry, as we ally for the development, acting as a development partner at scale.”

Revealed at CES was Intel’s Adaptive Control Unit (ACU), which claims to consolidate the control of vehicle domains, applications, and safety-critical functions all onto one central processing chip. 

More than just OTA updates, Intel simplifies hardware

Most modern vehicles employ a number of separate controllers, each with their own wiring and electronic control unit (ECU), either as part of a distributed approach or a consolidated zonal approach. Intel claims that with the ACU it can reduce overall energy demand from an EV’s battery pack by actually adapting the voltage level for high-voltage systems in real time, depending on the conditions (like driving style).

Its method, with a software-defined zonal controller, can freely shift workloads around. Tesla’s Sentry mode for vehicle security, for example, runs on the in-cabin system today and uses 40 to 50 watts constantly, mentioned Jack Weast, Intel’s VP and the general manager of Intel Automotive, in the company’s CES presentation earlier this month. But with a zonal platform it can run on just a few watts, turning on the larger system when there’s an intruder. 

Or, noted Weast later, if you’re at a long stoplight, some aspects of active-safety systems might not need to be powered up.

“The way that we look at it is to think about a data center; you get all these massive racks of computing but you can literally use software to activate one rack or the other, move workloads back and forth,” explained McCammon to Green Car Reports as we took in the show.

The Karma CEO noted the parallels between the idea and one industry standard that started with Intel and evolved from not just communication to scalable power and voltage—that of the USB port.

This sort of dynamic variable voltage scaling might have been used on portions of the EV platform previously, but expanding the concept to the entire vehicle is something new.

“This is what we’ve been doing in PC platforms for 20 years,” summed Weast in the CES presentation.

It would be a welcome change at this point. Today’s vehicles can be nightmares to service, let alone assemble—essentially rats’ nests of disparate control units, located around the vehicle and often running at different voltages and levels of connectivity, leaving some systems capable of over-the-air updates and others not. More than 50 ECUs isn’t uncommon in new vehicles, and some vehicles have more than 100. 

Intel’s solution could support “multiple topologies,” as part of a whole-vehicle architecture, it claims, and it could be the next step—or one of the next steps—in a quiet revolution in how vehicles are conceived, built, and upgraded. 

Building on Tesla and Rivian, or leapfrogging it?

Industry suppliers have some catching-up, measuring-up, and coordinating on standards to do, while there’s certainly potential to innovate. But some of this responsibility has been in the hands of automakers.

Shifting to a wiring-saving “zonal” architecture, with zones based on physical location rather than specific tasks, is one of the keys. So is shifting to software controls for those zones and the components within them. 

Tesla led more than a decade ago by applying the software-based, over-the-air-update model of smartphones to its vehicles—and applying the zonal approach—converging on the idea of systems that could be readily upgraded or features that might be enabled via software. 

Nio, among others, has also used a zonal approach. Rivian remains an industry darling of this transformation, and it managed in a rethink of its architecture that arrived last year in the 2025 Rivian R1S and R1T to consolidate an already-neat 17 ECUs down to just seven, cutting 1.6 miles of copper wiring from the vehicle, helping optimize efficiency and simplifying the supply chain. 

Under VW’s investment in Rivian, it will license that zonal design and allow future collaboration between the two companies. The approach could lead to more easily upgradable vehicles that might cost less to build and also might better cope with hardware changes over time. 

Although there’s no complete production vehicle with Intel’s approach yet, it claims an 80% reduction in ECUs, a 60% reduction in wiring harness length, and a 35% or more reduction in power consumption. Further, Intel says that the approach means lower vehicle component costs, allowing a smaller motor and battery pack—and lower-cost EV platforms overall. 

All this said, it’s unclear whether Intel’s solution is merely treading water among tomorrow’s EVs or swimming laps around them. Nearly every automaker by now understands the necessity to simplify in order to compete. Ford CEO Jim Farley, for instance, said in 2023 that its next-generation EVs will be “radically simplified,” with a new electrical architecture that’s fully software updatable. 

Karma’s EV inverter is a key piece

Back to the energy advantages, Intel says that its approach, including a power management system-on-chip (SoC), helps maximize efficiency for inverters, chargers, and converters. System-wide, the approach recovers up to 40% of what would otherwise be powertrain-system energy losses—netting a 3-5% gain in efficiency and range, by WLTP-cycle testing, along with faster charging and “a more responsive driving experience.” 

The inverter is one of the keys to that, and California’s Karma Automotive developed an inverter unit that’s intended to be part of Intel’s whole-vehicle solution. The silicon carbide inverter is Karma’s own intellectual property, confirmed Marques McCammon to Green Car Reports, and likely to be made in Michigan, but it’s co-branded with Intel.

 

By now you might be wondering: Why is the CEO of the California-based niche, low-volume luxury vehicle maker, which is best known for reviving and refining vehicles and systems derived from the first Fisker Automotive, talking industry-altering tech in such a deep way?

It stems from a long partnership when he was a management counterpart at Wind River systems—McCammon on the software side, Weast on the hardware side. 

“We had a vision then of what a software-enabled, or software-defined vehicle was going to be,” McCammon explained to GCR. “But when we thought this up, we always needed someone to go first.”

Fast forward a decade, to about a year ago, and Intel had just acquired Silicon Mobility, which McCammon said was a turning point. 

“We said, where do we start? Everybody’s focused on the cabin, but really energy management, usage, and efficiency is in the powertrain domains.” 

Karma had already been developing its own inverter, its own powertrain domain control unit, and a chipset, McCammon explained, but Intel and Silicon Mobility brought to the table “added a level of software programmability into a system that’s usually very fixed.”

McCammon explained that Karma essentially now it has a programmable API between motor and hardware—something most other modern EVs don’t have. “We were already working on it for our next inverter, for our next generation products,” he said. “It was a natural fit.”

“The inverter is one piece of it—this part we’re talking about today— but we are working on much, much more,” added McCammon.