The Lidar Quarterly Report: Q4 2025

Innoviz, Microvision, Aeva, AEYE

pre note: I’m bringing the em-dash back, I like them. Of course AI helped me write.

The Shakeout Is No Longer Theoretical

If you’ve been following these reports for 2025, you know the drill: I track (mostly) Western lidar pure-plays through their earnings, compare notes, and try to figure out who’s building a real business and who’s running out of time.

This quarter, the answer got a lot clearer.

Luminar filed for Chapter 11 in December. The highest-profile lidar company — the one with the Volvo partnership, the celebrity CEO, the $10B peak market cap — burned through $4 billion and sold its carcass to MicroVision for $33 million. Pennies on the dollar doesn’t do the math justice. This is pennies on the hundred-dollar bill. Now Microvision is touting that they have the breadth of offerings for anyone and everyone in the lidar space - because lack of focus has helped SO many companies reach profitability in this space.

Meanwhile, Ouster posted its first-ever GAAP-profitable quarter. Hesai shipped its millionth sensor and keeps printing money in China, and now with Mercedes perhaps outside of China. Everyone else scrambled to rebrand themselves as “Physical AI” companies, because apparently that’s what you do when automotive lidar timelines keep slipping to the right — You follow Nvidia’s lead.

Five companies now remain: Ouster (OUST), Aeva (AEVA), Innoviz (INVZ), Aeye (LIDR) and MicroVision (MVIS).

Five themes dominated Q4. Let’s get into it.

---

Four Themes from Q4 2025

Theme 1: Consolidation Reshapes the Landscape

The lidar sector finally got its reckoning.

Luminar’s Chapter 11 filing in December was the highest-profile failure in the space. Austin Russell’s company raised billions, signed up Volvo, Mercedes, and others, and still couldn’t turn any of it into a sustainable business. The assets — Iris and Halo sensors, roughly 30 customer relationships, and a pile of IP — went to MicroVision for $33 million. For context, Luminar’s cumulative net losses exceeded $4 billion. That’s a 99.2% discount on invested capital.

But Luminar wasn’t the only M&A story. Ouster scooped up Stereolabs for $35 million plus equity, adding 3D computer vision and software capabilities to build what CEO Angus Pacala calls a “Physical AI sensing platform.” It’s the most strategically coherent acquisition in the space — Ouster already had the hardware installed base, and Stereolabs gives them the perception software stack. Innoviz also launched InnovizThree, which has a RGB aspect to it — adding to the sensor fusion stack.

Aeva went the partnership route instead of acquisition, inking a strategic collaboration with LG Innotek worth approximately $50 million. LG gets to co-develop and manufacture Aeva’s FMCW sensors, and Aeva gets access to LG’s massive consumer electronics supply chain. Smart move — why build your own manufacturing when you can piggyback on one of the world’s best contract manufacturers? FMCW manufacturing has been a key challenge for the entire industry, since 2016. Perhpas this is the partnership that alleviates that.

And then there’s Hesai, the counter-narrative to all of this. While Western companies were acquiring each other’s scraps and announcing “strategic collaborations,” Hesai crossed one million cumulative sensors shipped and remained profitable. Dominant in China, increasingly competitive globally, and operating at a scale that makes every Western lidar company look like a rounding error.

The net result: fewer Western pure-play survivors. Less competition, sure — but also less sector-level investor interest. When your sector’s most famous company goes bankrupt, it’s hard to attract generalist capital.

Theme 2: The NRE-to-Product Revenue Transition

NRE — non-recurring engineering revenue — is essentially development fees that OEMs pay lidar companies to customize sensors for specific vehicle programs. It’s real money, but it’s not a product business. It’s consulting with hardware characteristics. Innoviz derives roughly 70% of its revenue from NRE. Aeva’s revenue mix is heavily NRE-weighted. The critical question for both: when does that NRE convert to recurring product revenue at scale?

The honest answer is: not soon. SOP (start of production) timelines for most automotive programs remain in the 2028–2031 window. A few years ago it was in the 2025 - 2027 window, and then that shifted to 2027 - 2028, and now… even further out. That’s three to five more years of burning cash before the recurring revenue engine kicks in — assuming programs don’t get cancelled, delayed, or re-competed. The only real programs set to launch soon, with any meaningful volume is VW’s Audi brand in H1 2027, and let’s assume that volumes don’t start high from the get go.

Ouster is the clear outlier here. With $146.6 million in product revenue and 25,000+ sensors shipped in FY2025, they have a real product business today. It’s not automotive — it’s industrial, smart infrastructure, robotics, and mapping — but it’s product revenue that ships, gets invoiced, and recurs. That matters enormously when everyone else is living on NRE promises. It’s also the reason that ALL the other companies are looking to enter the ‘physical AI’ space (see next point).

Theme 3: “Physical AI” as the New Narrative

“Physical AI” is in. On every Q4 earnings call.

Ouster coined the “Physical AI sensing platform” branding with the Stereolabs acquisition — cameras plus lidar plus perception software, all in one stack aimed at robots, autonomous vehicles, and industrial automation. Aeva followed with “Physical AI perception platform,” leaning into their FMCW technology’s velocity data as a differentiator for NVIDIA’s robotics ecosystem. Innoviz launched a “Physical AI” non-automotive revenue target, aiming for up to 10% of 2026 revenue from robotics and industrial applications.

Everyone is pivoting to the same story at the same time. Is that suspicious or simply the right business move?

Here’s the charitable read: the robotics and industrial autonomy markets are genuinely growing, lidar is a natural fit, and diversifying away from automotive-only reduces concentration risk. All true. Omer Keilaf, Innoviz’s CEO, has also stated a few times that the barriers to entry and speed to market for these segments is much much better.

Here’s the skeptical read: automotive SOP timelines are slipping, investors are losing patience, and “Physical AI” is a convenient rebranding that sounds sexier than “our automotive revenue is still years away, so we’re selling sensors to warehouse robots in the meantime.” A key question is that this was an industry built around the volume and revenue that the automotive sector provides - not warehouses or security.

The truth is probably somewhere in between. Ouster has the most substance behind the narrative — they’re already shipping thousands of sensors into these markets. Aeva has the NVIDIA reference platform and LG Innotek manufacturing partnership to back it up. For Innoviz and MicroVision, it’s still mostly PowerPoint, although Innoviz has shipped some early wins and to be honest - the use cases are so clear and simple that I honestly struggle to not see this pick up somewhat.

FYI I’m lumping defense into this category as well.

Theme 4: Cash Runway — The Survival Clock

Luminar’s bankruptcy made one thing viscerally clear: cash runway is existential. It doesn’t matter how good your technology is or how many design wins you’ve announced if you run out of money before production starts.

Here’s where everyone stands:

Ouster is in the best position — they just posted a profitable quarter, they have no debt, and their burn rate is declining. Three-plus years of runway with a realistic path to sustained profitability.

Innoviz has more than the 18 months of runway listed - they still have ~$50m of NRE’s coming in over the next two years, and a large portion of their ATM facility - meaning constant dilution - but no risk of going out of business.

Aeva has the most absolute cash ($247M including their credit facility), but they also burn the most ($120M/year). Two-plus years sounds comfortable until you realize that Daimler Truck production doesn’t start until 2028 at the earliest. The math gets tight. Especially if they don’t deliver their C sample in 2026.

MicroVision is the most precarious. Roughly 12 months of runway with $65–70M in guided operating expenses, a Nasdaq compliance warning to deal with, and going concern language in their filings. They need a capital raise, and they need it to work. They also need focus and slimming down - something they just did the opposite of.