AMPX — Amprius Technologies
The battery story is improving while my entry looks terrible. The latest news makes the commercialization case stronger, not risk-free.
Amprius makes silicon-anode batteries for places where weight really matters: drones, aviation, defense and selected mobility. This used to be much easier to dismiss as cool battery chemistry plus a PowerPoint. It is getting harder to do that now. Revenue is scaling, gross margin is improving, real customers are ordering, SiCore 500 gets to 500 Wh/kg on conventional manufacturing equipment, and the U.S. award helps fund domestic scale. Cool. Now I need those wins to become repeatable, profitable volume without dilution quietly doing half the work.
Amprius is still one of the more annoying positions I own.
Why I care about this one
The original thesis was simple: silicon-anode batteries can matter a lot in weight-sensitive applications if Amprius can actually manufacture them.
The second part is finally getting evidence behind it.
Amprius now has access to more than 2 GWh of annual production through its contract-manufacturing network, a multi-year Stark Future supply agreement beginning in 2027 with a total revenue opportunity above $100M, a $24M order from a European drone developer, cells in Redwire’s Stalker Block 30, and cells already flying in Matternet’s M2 platform. (Q2 results; Matternet)
This is where I need to not get excited and accidentally make the model stupid.
The >$100M Stark figure is a multi-year revenue opportunity. It is not $100M of 2027 revenue.
And the roughly $500M of new U.S. military orders management highlighted earlier this year were orders won by Amprius customers across defense programs. They improve visibility into future battery demand, but they are not $500M of Amprius backlog. (Q1 results)
I care about that distinction because this is exactly the kind of company where a giant TAM number can make a model stupid very quickly.
In September Amprius introduced SiCore 500, a second-generation cell that reaches 500 Wh/kg at a 1C continuous discharge rate and can be produced on conventional lithium-ion manufacturing equipment. Commercial availability is expected in Q4 2026. (SiCore 500)
The conventional-equipment part matters almost as much as the headline energy density.
If a great cell requires some bespoke manufacturing process nobody can scale, I do not care how impressive the lab result looks.
If Amprius can get that performance through existing manufacturing infrastructure, the route from chemistry to actual volume gets much easier.
Still needs to happen.
But that is real de-risking.
Q2 2026 + Sep. 2026 commercialization update
Primary sources: Q2 2026 results ↗ · Sep. 28 domestic production award ↗ · SiCore 500 launch ↗ · Q1 2026 results ↗ · Matternet partnership ↗
The $7 / $19 / $34 end-2027 centers stay put, but the model is now explicit: 2027 revenue × EV/Sales + net cash ÷ diluted shares. The operating evidence got better too. SiCore 500 reaches 500 Wh/kg on conventional manufacturing equipment, and the Sep. 23 U.S. award provides up to ~$75M toward domestic high-volume production with a 12M-cell annual line targeted for early 2028. That reduces financing and capacity risk. Because the line comes after the model horizon, I am raising confidence, not mechanically raising targets.
My position
What I actually bought and sold.
| Date | Type | Account | Dollars | Shares | Price | Return now* |
|---|---|---|---|---|---|---|
| May 19, 2026 | Buy | Taxable | $3,141 | 202.876 | $15.48 | -37.0% |
| Jun 25, 2026 | Buy | Taxable | $478 | 37.901 | $12.61 | -22.7% |
| Jul 14, 2026 | Buy | Taxable | $500 | 43.758 | $11.43 | -14.7% |
| Aug 20, 2026 | Buy | Taxable | $1,005 | 99.727 | $10.08 | -3.2% |
| Aug 31, 2026 | Buy | Taxable | $1,000 | 102.616 | $9.75 | +0.1% |
*Return now uses the latest available quote for buys and reinvestments. Sales and fees show “—”.
Valuation & scenario math
This was the part of the old page I did not like.
I had the targets. I did not show enough of the bridge underneath them.
I am keeping the end-of-2027 centers the same, but the framework is now explicit:
2027 revenue × EV/Sales multiple + year-end net cash ÷ diluted shares = value per share
These are my assumptions, not company guidance.
| End-2027 case | Bear | Base | Bull |
|---|---|---|---|
| 2027 revenue | ~$205M | ~$260M | ~$360M |
| Growth vs. $140M FY26 floor | ~46% | ~86% | ~157% |
| EV / Sales | ~5.4x | ~11.1x | ~13.9x |
| Implied enterprise value | ~$1.11B | ~$2.89B | ~$5.00B |
| Year-end net cash | ~$20M | ~$60M | ~$90M |
| Equity value | ~$1.13B | ~$2.95B | ~$5.09B |
| Diluted shares | ~160M | ~155M | ~150M |
| Value per share | ~$7 | ~$19 | ~$34 |
Bear math
$205M × 5.4x = ~$1.11B EV
Add roughly $20M of net cash:
~$1.13B equity value
Divide by roughly 160M diluted shares:
~$7.00/share
That is the “commercialization continues, but the market stops paying a heroic multiple for it” case.
Revenue still grows.
The stock can still disappoint.
Commercialization + manufacturing proof + financing discipline
No static price band here. The $75M award and SiCore 500 are real de-risking. Another dollar still needs repeat revenue, margin durability and manufacturing execution. The stock being below my basis is not evidence.
This is an end-2027 commercialization checkpoint, not a random target-price exercise. Each case is 2027 revenue × EV/Sales + year-end net cash ÷ diluted shares. The domestic-production award improves the financing/capacity side of the story. It does not magically turn into 2027 revenue because I want the target higher.
| Case | What has to be true | Target range | Annualized return from today |
|---|---|---|---|
| Bear | ~$205M 2027 revenue × ~5.4x EV/Sales = ~$1.11B EV; + ~$20M net cash = ~$1.13B equity value; ÷ ~160M diluted shares ≈ $7/share. Commercialization continues but growth slows, margins wobble and dilution remains part of the story. | $5.50–$8.50 center ~$7.00 | -23.8% |
| Base | ~$260M 2027 revenue × ~11.1x EV/Sales = ~$2.89B EV; + ~$60M net cash = ~$2.95B equity value; ÷ ~155M diluted shares ≈ $19/share. Revenue nearly doubles from the FY26 floor while gross margin and customer validation keep improving. | $17.00–$21 center ~$19.00 | +72.9% |
| Bull | ~$360M 2027 revenue × ~13.9x EV/Sales = ~$5.00B EV; + ~$90M net cash = ~$5.09B equity value; ÷ ~150M diluted shares ≈ $34/share. Multiple aviation/defense/e-mobility programs ramp together, SiCore 500 validates the technology at scale and financing becomes much less dilutive. | $29–$39 center ~$34 | +178.8% |
Returns use the current quote and the time remaining to Dec 31, 2027. The scenario targets stay fixed until the underwriting changes. For a short checkpoint, annualized returns can look extreme; the operating milestones matter more than the normalization.
Capital allocation
The grant helps a lot. Dilution did not suddenly stop existing.
Revenue, margin, customer validation and product performance are all moving the right way. The up-to-$75M milestone-based award makes the financing path meaningfully better. I still take points off for historical share growth, execution across multiple contract manufacturers and the chance commercialization eats more working capital than the headline EBITDA number makes obvious.
Stage-relative, not some objective rating-agency grade. The number is there to force the judgment into the open. See the rubric →
What management is actually doing with the cash
The domestic-production award is a real improvement. Up to about $75M, paid against milestones, to help convert an existing U.S. EV battery line into high-energy-density production with 12M-cell annual capacity planned by early 2028. That is way better than trying to fund the whole thing with equity. It is also not free money: the broader project is expected to cost about $100M including in-kind contributions, and the award is tied to the project. I still want capacity following actual customer demand, not TAM slides.
What I’m doing with the next dollar
$260M × 11.1x = ~$2.89B EV
Add roughly $60M of net cash:
~$2.95B equity value
Divide by roughly 155M diluted shares:
~$19.00/share
This is the case where the current customer wins start turning into repeat revenue, gross margin stays healthy, the company moves further into positive EBITDA, and the financing path looks meaningfully better because the government is helping fund domestic scale.
The part I would challenge hardest here is the 11.1x sales multiple.
That is expensive.
AMPX already embeds a serious forward-sales valuation depending on which revenue checkpoint you use. The operating milestones have to do the work from here.
So the base case is not “revenue grows and everything else stays the same.”
It requires the business to earn a premium valuation through execution.
What can break this
This is the update I care about most.
On Sep. 23, Amprius entered into a U.S. government agreement worth up to about $75M to establish domestic high-volume production for advanced silicon-anode cells used in small unmanned aircraft. The initial funded amount is $22.3M, payments are milestone based, and the project is intended to convert an existing U.S. EV battery line operated by a South Korean manufacturing partner. The completed line is expected to support about 12M cells per year and is targeted for early 2028. (award announcement)
For shareholders, that is a much better financing setup than ‘issue a pile of stock and hope.’
The project is still expected to cost roughly $100M including in-kind contributions from Amprius and its partner, so no, this is not a free $75M sitting in the bank.
But it does mean Amprius does not have to fund the entire domestic-production step with equity.
For a company where dilution is one of the main risks, that matters.
I moved the Capital Allocation score from 7.5 to 7.75/10 mostly because the financing path improved.
Not because the execution risk disappeared.
$360M × 13.9x = ~$5.00B EV
Add roughly $90M of net cash:
~$5.09B equity value
Divide by roughly 150M diluted shares:
~$34.00/share
That requires a lot to go right.
The Stark program ramps well.
Drone and defense programs become repeat orders.
Matternet and other aviation customers turn qualification into volume.
SiCore 500 is commercially validated.
Margins keep improving.
And Amprius scales without needing another ugly equity raise.
That is why it is a bull case.
Not because I can type “$34” into a spreadsheet.
The evidence I’m watching
Stark, the $24M drone order, Matternet/Redwire and the defense demand signals turn into repeat business; SiCore 500 actually ships commercially in Q4 and works on conventional lines; gross margin stays in the high-20s or better while volume scales; and the domestic build hits milestones without a meaningful new equity raise.
The wins stay one-off orders, Stark ramps slower than expected, SiCore 500 looks great in specs but does not get adopted, contract-manufacturing yields or margins disappoint, the domestic line slips beyond early 2028, or working capital forces dilution anyway.