Aeva's 2027 Optical Connectivity Play Could Scale to Millions of Units-If the Ramp Holds

Aeva's optical connectivity roadmap targets a much larger market than its current sensor business

Aeva is aiming at a market that could be much larger than its current sensor footprint. On an AI data-center optics roadmap, management has pointed to a potential multiple millions of units per year, with initial volume shipments targeted for 2027 and a ramp into 2028. For investors, the near-term question is whether this roadmap can move from timeline to proof.

The setup in one look

On paper, the opportunity is concrete. Aeva says it has signed a joint development agreement with a major customer for a near-package optics solution tied to a hyperscaler deployment, and it launched a dedicated optical connectivity business to commercialize the effort. If the company can translate its laser and silicon photonics expertise into a qualified supply relationship inside hyperscaler infrastructure, today's small revenue base would be best viewed as an early stage rather than the final conclusion.

Why skeptics still have a case

The caution is straightforward: small numbers can still reflect real execution risk. Aeva posted $6.1 million of Q2 2026 revenue, but that growth came entirely from professional services, while the broader business still faces delayed automotive ramps and scaling pressure. Bulls can argue a new revenue engine may yet emerge. Bears can argue the core product pipeline is not yet proven. The next few quarters should clarify which view is more reasonable.

AI data-center design is starting to run into copper limits

The market need is no longer just "more bandwidth." Industry analysis describes an "interconnect wall" as switch IC bandwidth continues to advance. In practical terms, the chip is getting faster than copper links can comfortably carry data over any meaningful distance. That makes optical interconnects look less like a optional upgrade and more like a possible infrastructure requirement.

Where copper starts to strain

The bottleneck shows up in several places at once. The same report points to electrical signaling limits, front-panel density constraints, and rising power consumption for longer electrical links. It also flags broader data-center pressures around thermal management, cable management, reliability, and serviceability.

That is why optics matter now. Shortening the link between the switch or compute chip and fiber can reduce the distance electrical signals have to travel. The goal is not simply higher speed on paper; it is a rack design that remains usable as power, cooling, and density pressures rise.

Why NPO and CPO matter now

Near-Package Optics and Co-Packaged Optics both move the optical interface closer to the chip, which is where the industry needs relief if switch ASIC bandwidth keeps outrunning copper's practical limits. For investors, that matters because packaging shifts usually create value in specific component slices, not in some vague "optics will grow" way.

Aeva is targeting one of those slices. It is focused on high-power optical source technology for next-generation optical engines in near-package and co-packaged optics designs. Just as important, this is not only a slide opportunity: the company has also signed a joint development agreement with a leading high-speed optical engine provider for an NPO solution tied to a major hyperscaler program.

If the market keeps moving toward package-close optics, being inside that supply chain before the category fully standardizes could matter far more than today's revenue base. The key boundary condition is reliability: the product still has to clear the durability and serviceability expectations of real data-center hardware.

Aeva still has to prove it can scale product demand and fund the build-out

The real test is not whether the optics opportunity looks big on a roadmap. It is whether Aeva can build repeatable product demand now and still have enough cash to reach the next milestone.

Production progress is real, but revenue mix remains the watchpoint

There is at least one concrete operating track record already under way. Aeva and Daimler Truck plan to expand series-production LiDAR capacity of up to 200,000 LiDAR units annually, and the first autonomous-ready Cascadia trucks equipped with Aeva sensors are now operating on Texas roads. That matters because it shows the company is not stuck in lab demos. It has a production roadmap, a serious automotive relationship, and hardware that is already in use.

But investors still need to see that progress turn into durable product revenue. In Q2, professional services accounted for approximately 59% of quarterly revenue. Bulls can argue services are the bridge while product volumes catch up. Bears will argue the harder conversion has not happened yet: the higher-quality, scalable revenue stream is still not the bigger one.

Cash burn keeps the timeline under pressure

Funding is where execution risk becomes most visible. Aeva used $57.0 million of cash in operating activities in H1 2026, a meaningful drain for a company that is still proving it can convert pipelines into repeatable shipped units. DTNA's additional non-recurring funding helps, but one customer support package does not remove the need to build a broader revenue base.

The main proof points to watch are straightforward:

• Product mix improves, with product revenue overtaking services.
• Automotive deliveries scale without another soft ramp.
• Cash use slows enough to give the company more time before funding becomes the headline.

What would make the optical connectivity story more credible?

For now, this is not a finished AI infrastructure story. It is a real roadmap with a funding question attached, and the most reasonable stance is selective, evidence-driven watching.

What deserves a second look

A second look becomes more compelling when the narrative stops being mainly about future scale and starts showing near-term proof. The clearest upside trigger would be progress on the joint development agreement that moves from concept work into named-customer validation, design wins, or production milestones. Investors should also watch whether the business mix improves away from a period when professional services accounted for approximately 59% of quarterly revenue. If the project-style business remains the larger line item while core product conversion lags, the market has less reason to pay up.

Catalyst watchlist

• Named-customer or production proof in the optical connectivity program, beyond general validation language.
• A clearer bridge from development to shipped units, so investors can see repeatable demand rather than only partnership headlines.
• Some relief on execution pressure in the broader business, including delayed automotive ramps and scaling challenges.

What would weaken the setup

If cash burn keeps outrunning product conversion, or if 2027 remains a timeline without intermediate proof, this stays more of a hope story than an investable ramp. Optical connectivity remains option value until development progress turns into named-customer or production evidence, and the first concrete proof point will matter more than the bigger prize on the slide.