Battery Test Worth Billions

🌾 Welcome to StableBread's Newsletter!

Phones, EVs, AR headsets, drones, and humanoid robots all need to store more energy in less space, and the standard lithium-ion battery design hasn't structurally changed since Sony commercialized it in 1991.

Enovix (Nasdaq: ENVX) is one of the few companies innovating a new physical battery design, manufactured in its own automated plant, and is now in qualification testing with the largest smartphone makers in the world.

The stock trades around $5.22/share, down ~63% over the past year:

The drawdown is largely due to a single battery test the company has yet to pass. Whether that test gets resolved decides most of what this $1.1B company is worth.

Below, I'll walk through the business, the qualification mess, the financials, the insider selling, valuation, and whether I’m buying at these prices.

Enovix designs and manufactures lithium-ion batteries whose anode is made of 100% active silicon. The “anode” is the battery's negative electrode, the side that stores lithium ions when you charge, and nearly every battery in your phone, laptop, and earbuds uses graphite instead.

Silicon can theoretically store more than twice as much lithium as graphite, and the battery industry has pursued it as an anode material for decades. The problem is that silicon swells up to 60% when charged, then shrinks when discharged, cracking the material and killing the battery within a handful of cycles.

That's why incumbents like ATL (Amperex Technology), Samsung SDI, and LG Energy Solution only blend small amounts of silicon (5-15%) into their graphite anodes.

Enovix's solution is architectural rather than chemical. Instead of winding long electrode strips into a "jelly roll" like a conventional cell, it stacks many short anodes and cathodes side by side, then locks the stack inside a stainless steel case (Enovix calls it the "constraint system") that squeezes the silicon so it can't crack apart. An extra lithium layer ("pre-lithiation") replaces the lithium that silicon permanently traps on first charge.

That architecture is what produces the energy density. In January, an independent lab measured Enovix's AI-1 smartphone battery at 935 Wh/L—12% higher than the best competing battery tested using the exact same methodology, which was a conventional cell with a graphite-silicon blend anode.

Enovix's chairman, T.J. Rodgers, frames that 12% as roughly a two-year lead at the industry's historical improvement rate of 7% per year.

Enovix was founded in 2006 by veterans of the hard-disk-drive and semiconductor industries. Rodgers, the founder and longtime CEO of Cypress Semiconductor, joined as an investor and board member in 2012 and took the company public through his own SPAC in July 2021.

His quarterly letters read more like an engineer's notebook than investor relations copy. In the Q1 2026 letter he called the Enovix battery "the single most difficult project I've ever worked on," and admitted the company's original concept (building batteries inside grooves etched into silicon wafers) never worked at all.

The stacked-electrode design described above was a last-ditch pivot, made when the founders were down to their final $200,000 of venture funding.

Raj Talluri, who previously ran Micron's mobile business, became CEO in early 2023. Raj reshaped Enovix by dropping the old strategy of serving hundreds of small IoT customers with standard cells and going after a few large customers with custom designs, starting with smartphones, the most demanding battery market in consumer electronics.

He also moved manufacturing to Asia. Fab1 in Fremont was shut down (a $41.8M restructuring charge in 2024), a high-volume plant called Fab2 opened in Penang, Malaysia, and Enovix acquired Routejade, a 20-year-old Korean battery maker, in October 2023.

Today the company sells two product platforms out of this footprint (market sizes based on company estimates):

You'll notice robotics on the long-term target list for both platforms. Management named humanoid robots as a future application for the 100% silicon platform on the May earnings call, alongside EVs and computing, and robotics leads the MX platform's longer-term roadmap next to eVTOL and medical devices.

There are no robot contracts, customers, or revenue today, and Enovix hasn't announced any. The long-term logic is that a humanoid robot is a battery-constrained device like a phone or a drone, just bigger, and the energy density that wins the small form factors is the same edge that would matter there. I'd treat it as a nice bonus that only exists if the phone and drone businesses work first.

Enovix also owns its factories in Malaysia and Korea rather than using Chinese contract manufacturers. Management says this matters to both defense customers (who can't buy Chinese cells) and consumer OEMs diversifying supply chains away from China. And since no contract manufacturer takes a cut, unit economics should scale better if volume arrives.

Enovix signed a development agreement with a leading smartphone OEM in September 2024. On the Q4 2025 call, management named the customer as Honor, one of the top Chinese smartphone makers, and said it remains engaged with seven of the top eight global smartphone OEMs, with a second unnamed Asian OEM close behind Honor.

Management repeated the point in May, adding that it's in active dialogue with several more leading OEMs.

The original plan, per a March 2025 press release, was mass production in late 2025. That slipped to late 2026. The current plan, per the Q1 2026 report, is a small "system-level deployment" in the second half of 2026, with a broader commercial launch in 2027.

Two years of slippage on its most important product is the primary reason the market has lost patience.

What's Actually Holding It Up

Honor's qualification process spans 75 specifications. Per Rodgers, Enovix has passed 72 of them, up from 70 a quarter earlier. Energy density, fast charge, and safety are done. The three remaining are two cycle-life tests and one below-freezing power test.

Cycle life is measured at different discharge speeds, called C-rates. A 1C rate drains the battery in one hour, and a 0.2C rate drains it in five hours. Honor's primary spec is 1,000 charge cycles at 0.2C, which reflects how phones are actually used. Enovix's internal data shows its cells on track to exceed that.

The problem is that running a true 0.2C test takes about a year, so the industry compresses testing by cycling batteries at an accelerated 0.7C rate, with a secondary spec of 800 cycles.

That 0.7C accelerated test was designed around graphite batteries, and silicon behaves differently under fast discharge. This isn't just management talking, as peer-reviewed battery research documents how higher C-rates accelerate degradation and expansion in silicon-based anodes.

While Enovix's cells routinely hit 500 cycles at that 0.7C rate, they struggle to reach the 800-cycle threshold.

Management's argument, laid out with its Q4 2025 results in February, is that 0.7C is faster than any real smartphone workload (even video recording only draws ~0.17C) and that the accelerated test punishes silicon in a way real usage doesn't.

In May, Enovix announced it won that argument. Honor agreed to remove the 0.7C test as a hard requirement and adopt a silicon-specific qualification framework built around 0.2C and 0.1C testing that mirrors real-world usage. The second Asian OEM mentioned earlier acknowledged the same flaw in the legacy test and is moving toward a similar framework.

Management says discussions with additional Tier-1 OEMs are moving the same way, and that this removes the "principal structural barrier" to qualifying 100% silicon batteries industry-wide.

How to Read This

To a skeptic, Enovix couldn't pass the test, so they rewrote the protocol. But regardless of the narrative, the stock fell 14% on the Q1 report simply because the new framework extends the testing window, pushing final qualification timing out yet again.

I land somewhere more charitable, for three reasons:

That said, charitable doesn't mean resolved. Cycle-life testing under the new framework is "approaching required thresholds," which is management-speak for not passed yet. The below-freezing power test (silicon delivers less instant power than graphite in extreme cold) is also unresolved.

Until Honor signs off, everything downstream, including the 2027 volume ramp, is a plan rather than a contract.

So why do the OEMs keep working with Enovix through the delays? Phone batteries are heading toward 10,000 mAh as on-device AI drains more power, phones can't physically get bigger, and energy density in 2024 flagship phones only improved ~1% y/y by Enovix's analysis.

Dropping in Enovix's battery would add 12% more energy in the same space, effectively packing a decade of industry advancement into a single upgrade.

If AI-1 ships in a flagship phone at scale, a single smartphone program fills Enovix's entire current Penang production line, per Rodgers, and a few programs fill the four lines the plant can house.

The smartphone battery is still stuck in testing, but the same silicon cells found their first paying market in smart eyewear.

In January 2025, Enovix announced a pre-paid purchase order from a Silicon Valley-based AI and immersive-technology company for batteries in head-worn mixed-reality wearables. Enovix hasn't named the customer.

By the Q1 2026 report, this had become a production order. Enovix began commercial manufacturing of AI-1 cells for the customer's smart eyewear reference platform, with ~50,000 units expected in 2026 and volume "in the millions" expected next year as more products launch.

Fifty thousand small cells won't move revenue much this year. What it proves is that the 100% silicon anode architecture can be manufactured at commercial scale, a first for the industry per management, and it puts Enovix inside the supply chain of smart eyewear, one of the fastest-growing consumer device categories.

Management expects smart eyewear unit shipments to grow more than 5x through 2030, with display-enabled glasses drawing 2.5x the power of today's displayless models. Enovix estimates the eyewear battery market alone could exceed $1B by 2030.

Eyewear is also where Enovix's design edge is largest. The steel case takes up proportionally less space in a small cell, so the smaller the battery, the bigger the energy density gap over conventional designs. And eyewear makers require far fewer charge cycles than smartphone makers do, so the qualification wall that blocks smartphones doesn't exist here.

The batteries also use 12x less raw material than a phone cell, so they run faster and yield better through the same production line.

Most smart glasses are built on a reference design, a ready-made hardware blueprint from a tech or chip company that consumer brands then wrap in their own frames (think Ray-Ban-style distribution).

Getting designed into one reference platform means Enovix's battery ships in every product built on it. Enovix has shipped eyewear samples to 15 customers, and its cells hold both CCC (China Compulsory Certification) and UL safety certifications.

Enovix is already onto the next cell. AI-2, built on the newer EX-3M technology node, delivers more than 20% higher volumetric energy density than AI-1 (420 mAh vs. 355 mAh in the same eyewear format) by cutting inactive material and raising cathode voltage from 4.43V to 4.52V.

Engineering samples were completed in Q1 and customer sampling began this quarter. The same improvements roll into future smartphone cells, which is how the technology lead compounds.

Nearly all of Enovix's actual revenue today comes from South Korea.

The Routejade acquisition in 2023 brought two factories in Nonsan City with four automated battery production lines, two electrode coating lines, and a 20-year history supplying Korean defense contractors.

In April 2025, Enovix bought an adjacent battery cell plant from SolarEdge for $10M, adding ~300,000 sq ft of floor space, and booked a $4.8M bargain purchase gain because the assets were worth more than it paid. Management says the combined footprint can support ~2x current revenue before major new investment.

These facilities make conventional graphite and silicon-doped cells for defense and industrial customers, including aerial drones, subsea systems, munitions, wearable soldier systems, and medical devices. Naval munitions batteries were the top product in Q4 2025. Enovix has supported US Army programs since 2021 and completed final deliveries under the Army's Conformal Wearable Battery program.

Two concentrations are worth flagging:

But these programs are long-term, deployed, and reordering (Q4 2025 set a revenue record on defense shipments), and the pipeline is diversifying fast.

Still, a customer this concentrated can make any quarter lumpy, and Q1 revenue landing below Q4 in each of the last two years reflects exactly that defense order seasonality.

The growth driver here is drones. Demand for Western, non-Chinese drone batteries currently exceeds available supply, and Enovix's global pipeline for Korea-manufactured products now exceeds $130M, up from ~$100M entering 2026, with over 60% of it drone-related.

That $130M is the lifetime value of new designs in progress, per management's definition, not annual revenue, and pipeline is not backlog. Conversion is what ultimately matters.

In May, Enovix formally launched its first drone product at the Michigan Defense Expo. The MX1-B01 cell delivers 360 Wh/kg of gravimetric energy density (energy per unit of weight, the metric that matters for flight time) with 5C continuous and 8C pulse discharge, and blends ~60% silicon-carbon into the anode.

Drones tolerate what smartphones can't. Some swelling is fine inside a drone bay, and 300 cycles is plenty for a drone's typical lifespan, allowing Enovix to trade cycle life for maximum energy. The MX-2 targets 400 Wh/kg in 2027.

The MX1 cell is also NDAA-compliant, meaning it satisfies the National Defense Authorization Act sourcing rules that restrict US defense buyers from using batteries made in, or sourced from, "foreign entities of concern" (FEOC), namely China.

Compliance requires both non-FEOC manufacturing (Nonsan qualifies) and non-FEOC material sourcing, which management says is hard for competitors relying on Chinese contract manufacturing to replicate.

Enovix plans to show MX-1 at 11 more defense conferences across the US and Europe over the next two quarters, and hired Steve Bakos, a 35-year semiconductor sales veteran from Infineon who managed large global accounts including Apple, as SVP of Worldwide Sales in May to convert this pipeline.

Enovix discloses more manufacturing detail than most pre-volume companies, which makes its progress checkable.

Fab2 in Penang runs three lines:

Overall line yield is the product of nine process steps. A year ago several steps were problems. As of the Q1 report, most zones are at or above 90% yield, with the remaining two around 80% and 88%.

The one stubborn constraint is Zone 1 laser dicing, the step that cuts electrode ribbons out of rolls of coated foil. The cobalt-oxide cathode material is hard, lasers cut it slowly, and each additional laser costs ~$1M.

Rodgers disclosed in his Q1 2026 shareholder letter that Enovix decided against buying the extra lasers needed to reach full line speed (1,350 units/hour) because the economics don't justify it. Instead, the company spent over a year developing a mechanical die-cutting alternative and showed a working prototype video on the May call.

The plan is a hybrid laser-plus-mechanical configuration, with the mechanical dicer online this year to support next year's volumes at lower cost.

Rodgers adds two caveats:

Capacity for 2026 demand is covered by the existing lasers, and 2027 is what the mechanical dicer is for.

Per the 2025 10-K, the Fab2 building lease expires in July 2026, with renewal discussions underway and management evaluating purchase options.

The July 9 announcement addressed who runs a multi-site, multi-product ramp. Enovix hired Dr. Michael Vyvoda as COO, effective July 29. Vyvoda spent 5+ years at Apple as Director of Product Operations for audio products, where he helped scale AirPods from new-product introduction to high-volume production across multiple Asian sites.

He inherits a unified operations organization spanning KH Park (global manufacturing), Ed Casey (advanced manufacturing engineering, ex-Seagate/Western Digital/ams OSRAM), and James Wilcox (strategic sourcing). Someone who scaled AirPods doesn't take this job unless he believes the ramp is real.

Enovix's reported numbers are small, but the direction is right on every line, including cash burn, which improved in 2025 as Fab2 construction spending rolled off.

The issue is size, as the company still burns ~$30M a quarter against $7.6M of revenue. I'll address whether that cash burn is manageable when we break down the balance sheet.

Start with the FY 2025 results:

Q1 2026 kept the pattern. Revenue of $7.6M beat guidance and grew 49% y/y, non-GAAP gross margin reached 26.3%, and it was the sixth consecutive quarter of positive gross profit on both GAAP and non-GAAP bases.

Q2 guidance calls for $8-9M of revenue, which includes the first smart eyewear deliveries, against a $29-32M non-GAAP operating loss.

Gross margin turned positive because Korean production scaled and the product mix shifted toward higher-margin defense cells, not because the silicon battery is profitable yet.

When the smartphone and eyewear production ramps, some costs now sitting in operating expenses will move up into COGS, so the reported margin path won't be a straight line.

Ryan Benton, the CFO, points out that materials make up most of cost of sales, and that some AI-2 roadmap changes cut the bill of materials while raising energy density. The newer cells should cost less to build and sell for more.

Operating expenses are the price of running a smartphone qualification, an eyewear ramp, a drone launch, and a factory scale-up simultaneously.

GAAP opex ran $45.4M in Q1 (roughly $26.5M R&D, $18.9M SG&A), and stock-based compensation (SBC) adds ~$49M a year, meaningful dilution against a $1.1B market cap.