5 in 5: Five Speculative Small-Caps with Five-Year Breakout Potential

The most useful small-cap wins rarely look obvious in advance. They look speculative, unfinished, and a little ridiculous right up until the moment execution catches up with the story.

AST SpaceMobile $ASTS ( ▼ 7.83% ) (which I first flagged in February 2024 at ~$3.30/share) spent years as exactly that kind of name, a plan to beam broadband from orbit straight to an unmodified smartphone with almost nothing operational to show for it.

Then the BlueBird satellites reached low Earth orbit, the carrier agreements with AT&T and Verizon turned into a commercial path, the spectrum rights stacked up, and in May 2026 the FCC cleared the company to run commercial service in the United States. It booked its first revenue in 2025 and now trades north of $25 billion. The speculative small-cap became the thing the speculation was about.

This screen groups five names that sit roughly where AST SpaceMobile sat a few years ago. Each is small, most are pre-profit or barely past it, and each is chasing a market large enough that hitting its development milestones would push it well out of small-cap range.

NANO Nuclear is a roughly $980M developer of microreactors, the shipping-container-scale fission units pitched as portable power for data centers, remote sites, and defense installations.

Its lead system, the KRONOS MMR, is advancing through the early regulatory pipeline, and the company is working to vertically integrate into the high-assay low-enriched uranium fuel that advanced reactors will need.

What it is not yet is a revenue-generating business. NANO is pre-commercial, and every number that matters is a projection.

What makes it a five-year candidate rather than a lottery ticket is the collision of two things: a real and growing power-demand problem, and a balance sheet that can fund a long wait. The AI build-out has turned electricity into the binding constraint on data-center growth, and microreactors are one of the few supply answers that do not depend on the existing grid. NANO ended its fiscal second quarter with roughly $569 million in cash and short-term investments against an effective $900 million shelf registration, giving it years of runway to reach a licensed design.

That runway is also the tell on the risk. A company with this much cash and no product is, by definition, spending against milestones it has not hit, and the microreactor field has a long history of timelines slipping to the right. The stock has already re-rated hard, up from under $150 million in 2024, so a good deal of the eventual success is priced in, and the share count keeps climbing as the shelf gets used.

The catalyst worth watching is narrow and specific: a licensed and buildable reactor design. Everything NANO is worth today rests on a reactor that does not yet legally exist.

CVRx is an outlier in speculative small-cap stocks with long-term upside as a roughly $114M medical-device company with a 0.78x beta. Its flagship product, Barostim, is the first neuromodulation device the FDA has cleared to treat heart-failure symptoms, an implant that sends electrical pulses to the baroreceptors in the carotid artery to rebalance an overtaxed autonomic nervous system.

Commercially it is early but working. First-quarter 2026 revenue rose about 20% to $14.8 million, U.S. revenue grew 22%, gross margin sat near 87%, and the company held about $72.3 million in cash across 257 active U.S. implanting centers.

Two things make the next five years interesting.

The first is reimbursement: the Category I CPT codes that took effect on January 1, 2026 replaced the old Category III codes that triggered automatic insurer denials, and CVRx's own data shows 30-day Medicare Advantage prior-authorization approvals climbing from 31% in 2024 to 50% in early 2026.

Quote: Kevin Hykes, President and Chief Executive Officer of CVRx. “We are also starting to observe positive effects from the Category I CPT Code that took effect at the beginning of the year, as early data shows improved 30-day prior authorization approval rates compared to what we saw in 2025.

The second is the BENEFIT-HF trial, a randomized study that, if it reads out well, could expand Barostim's indicated patient population roughly threefold.

The risk, as with many frontier-MedTech small-cap stocks, is patience. CVRx still spends far more than it earns, adoption has been slower than bulls hoped, and the trial that could triple the market will take years to complete. In other words, the base business funds that wait but the trial is the swing factor.

Palladyne AI, the roughly $250M company that emerged from the old Sarcos robotics business, sells the intelligence layer rather than the machine. Its SwarmOS software lets a single operator command a mixed fleet of drones that keep coordinating even when communications are jammed, running on the small compute onboard each aircraft rather than in the cloud. That is a genuinely hard problem, and Palladyne has spent the past year proving the software works outside a lab.

The proof points are specific:

The financials, however, point to the speculative nature of PDYN in a way that its consecutive wins conceal: preliminary second-quarter 2026 revenue was about $5.8 million, up roughly 480% year over year but off a tiny base, against a backlog near $24 million and cash of about $44 million.

Full-year guidance of $24 to $27 million implies a steep second-half ramp that has to actually arrive, and government contract timing is lumpy enough that a slow quarter is always possible. Against that cash balance, further dilution is the likely funding path.

The five-year question is whether SwarmOS becomes the autonomy standard that other people's drones adopt, similar to how Anduril’s Lattice is becoming an increasingly central command and control system for MilTech hardware providers.

If Palladyne is the brain, Draganfly is one of bodies it runs on. The two are not just thematically paired as DroneTech cousins: Draganfly has tested Palladyne's SwarmOS across its own unmanned platforms, which is the cleanest illustration of where the software layer meets the hardware layer in this cohort.

Draganfly is a Saskatoon-based drone builder with more than 25 years of history, valued around $175M (the company reports in Canadian dollars, so figures convert). It makes physical systems: the Flex FPV first-person-view drones, the Commander and Heavy Lift multirotors, fixed-wing aircraft picked up through its Skip Dynamix acquisition, and the a range of stabilized camera-and-sensor payloads, plus hardware payloads that include winching and delivery platforms.

The demand signal is the story, in that Draganfly's Flex FPV systems were selected by two separate U.S. Department of War units in May 2026, and the company was chosen alongside F4 Defense International for a DEVCOM Army Research Laboratory contract to develop a mobile counter-drone system.

Both slot into the same reality Ukraine made unavoidable: cheap attritable drones now shape battlefields, and Western militaries are scrambling for domestically produced supply.

First-quarter 2026 revenue was a record for the company at about $2.31 million, up roughly 49%, real growth on a very small base and nowhere near profitability. Draganfly has funded itself with repeated stock sales, so the balance sheet is well capitalized but the share count is a moving target, and the stock is violently volatile, with a whopping 3.7x beta.

The five-year question is whether early orders harden into programs of record. Pilot buys are cheap for a military to hand out but sustained procurement is the thing that would move this name into neo-Prime territory.

Serve Robotics may be the most recognizable idea on this list and, at roughly $450M, the one furthest along in deployment. Its sidewalk robots handle last-mile food delivery, and the pitch is straightforward arithmetic: the median U.S. food-delivery trip is about two and a half miles, and sending a two-ton car and a paid driver to move a burrito that far is an expensive way to do it.

Serve spun out of Uber in 2021, counts Uber and Nvidia among its strategic backers, and designs both the robot and the autonomy stack that runs it.

The scale is real, as Serve has roughly 2,000 robots operating across 44 cities in 14 states, has passed nearly two million cumulative deliveries, and plugs into both Uber Eats and DoorDash, which together cover most of the U.S. food-delivery market.

In 2026 it bought Diligent Robotics, pushing into indoor hospital robots and giving it a second vertical beyond the sidewalk in a sector that’s likely going to see sky-high growth in coming years (MedTech/healthcare robotics – more on this soon as I prep a robotics deep dive!).

First-quarter revenue hit $3.0 million, up almost sixfold year over year, and the company sits on about $197 million in liquidity.That liquidity is doing heavy lifting. Serve guides to roughly $26 million in 2026 revenue against operating expenses of $160 to $170 million, so it is spending many multiples of what it earns to build the fleet and the data flywheel behind it.

Whether that flywheel produces durable per-delivery profit is the entire question, and it has not been proven at scale yet. The number worth tracking over the next five years is not robot count, which Serve can always buy but instead margin per delivery, which it has to earn.

Hit reply with thoughts, corrections, or names worth a closer look. If you’re seeing this outside of your inbox, shoot me an email here.

Keep digging and see you next time!