Enovix (ENVX) 2026财年第二季度业绩电话会：智能眼镜业务开始贡献收入，储备项目规模达1.83亿美元

Enovix Corporation（NASDAQ：ENVX）公布2026财年第二季度营收为900万美元，主要受韩国国防出货量的支撑。该公司还首次录得智能眼镜产品收入，并推进了与核心客户的智能手机电池认证进程。

核心要点

• 2026财年第二季度营收同比增长21%，环比增长19%，达到900万美元，标志着Enovix连续第五个季度实现营收同比增长。
• 核心智能手机客户确认，Enovix电芯在0.2C放电循环测试下已通过超过1,000次循环。管理层预计剩余的加速循环寿命测试将在2026年底前完成。
• Enovix向某头部（Tier 1）智能眼镜客户出货了约2,100块AI-1电池，并确认了该项目的首笔产品收入。公司计划在第三季度交付约19,000个电池包。
• 韩国产品管线规模环比增长41%，预估年产值峰值约为1.83亿美元。仅无人机领域的业务机会就超过1亿美元。
• 截至本季度末，Enovix拥有的现金、现金等价物及可变现证券约为5.521亿美元，其中包括受限现金。
• 第三季度营收指引为900万至1,000万美元，非GAAP运营亏损为2,900万至3,200万美元。

关键财务业绩

季度毛利率下降主要反映了通过韩国业务销售的电池产品组合的变化。然而，上半年非GAAP毛利率从上年同期的21.3%上升至22.8%。此外，第二季度也是Enovix在GAAP和非GAAP准则下连续第七个季度实现毛利润为正。

业务与运营表现

智能手机认证取得进展

Enovix的核心智能手机客户独立确认，其电芯在0.2C放电协议下的循环次数已超过1,000次。剩余的认证工作涉及与客户共同开发的加速混合循环寿命协议，该协议用于替代传统石墨电池采用的0.7C测试。

管理层预计将在2026年完成最终测试。随后预计将进行小规模真机组装和实地测试，目标是在2027年实现量产。Enovix还预计将在2026年第四季度开始向第二家智能手机原始设备制造商（OEM）交付样品。

智能眼镜进入商业化量产

Enovix在第二季度出货了约2,100块AI-1电池，并确认了来自某头部客户的首笔智能眼镜产品收入。公司在第三季度拥有约19,000个电池包的交付订单，约为第二季度出货量的九倍。这些交付是一项50,000个电池包订单的一部分，剩余部分预计将在第四季度发货。

该公司还向一家头部智能眼镜客户提供了AI-2电芯样品。管理层预计，通过使用更薄的材料、提高封装效率以及更高的正极电压，AI-2的体积能量密度将比AI-1高出约20%。

无人机与国防业务管线扩大

韩国制造产品的管线规模从第一季度增长了41%，达到约1.83亿美元。Enovix将这一数字定义为预估年产值峰值，而非未完工订单（Backlog）。其中无人机领域的业务机会超过1亿美元，而整个管线业务中有超过4,000万美元涉及客户正在积极测试电芯或将其设计到产品中。

MX-1的目标能量密度约为每公斤360瓦时，同时支持高持续放电和脉冲放电。Enovix预计韩国新增产能将在2027年中期投入运营，随着客户项目完成认证，随后将进行首次商业出货。订购的设备预计将支持约100万套的年产能。

管理层继续将MX-2的目标定在2027年，目标是达到每公斤400瓦时。

制造良率提升，但切割仍是瓶颈

Fab2的累计良率连续第三个季度获得提升。在Zone 1区域之外，除一道工序外，所有工艺步骤的良率均达到至少95%，部分甚至达到99.6%。

Zone 1的切割良率从第一季度的80%提高到约84%，但仍是主要的吞吐量瓶颈。Enovix正在实施激光与机械混合切割配置，预计多道机械工序将在年底前后上线。

管理层业绩指引

营收指引范围代表约13%至25%的同比增长。管理层表示，季度营收和毛利率可能会随着产品组合、交付时机、认证进度以及商业化爬坡计划而有所波动。

风险与关注焦点

• 最终的加速智能手机循环寿命测试仍是在进行真机测试和商业推向市场之前的认证关口。
• 预计在2026年剩余时间内，智能眼镜生产的毛利率将为负，因为初始产量不足以吸收制造固定费用。
• 尽管良率有所提升，但Zone 1的切割仍是Fab2的主要吞吐量瓶颈。
• 韩国业务的毛利率可能会发生波动，因为不同的产品SKU具有不同的毛利率。
• 1.83亿美元的管线规模代表的是预估年产值峰值，而非确定的采购订单。转化情况取决于客户的测试、认证和方案中标（Design Win）。
• 根据管理层在问答环节中的表述，无人机公司可能会在大约六到九个月内迈向量产，而国防主承包商项目可能需要大约18个月。

分析师问答环节点评

管理层拒绝提供智能眼镜的具体盈亏平衡出货量或定价。管理层表示，目前的首要任务是扩大规模，将Enovix打造成常态化主导供应商，随着客户采纳度和出货量的增加，预计毛利率将趋于更健康的水平。

在智能手机方面，管理层表示，0.2C测试结果提供了经客户确认的循环寿命性能证明。剩余的混合加速测试旨在比传统的基于石墨的协议更准确地反映真实世界的手机放电模式。

对于无人机和国防领域，管理层表示需求可能会迅速超出当前供给。公司正在结合管线转化分阶段扩充韩国产能，预计新设备到2027年中期将可支持每年高达100万套的产能。

管理层还表示，HVM-1生产线仍保持每小时1,350套的标称产能，不过人员配备和生产节奏正在根据客户的认证时间表和需求进行调整。

财报电话会议完整文字记录

---

完整财报电话会议逐字稿

管理层陈述

Operator

Thank you for standing by, and welcome to the Enovix Corporation Second Quarter 2026 Earnings Conference Call. [Operator Instructions] As a reminder, today's program will be recorded.

And now I'd like to introduce your host for today's program, Monica Gould, Investor Relations for Enovix. Please go ahead.

Monica Gould

Thank you, operator. I would like to welcome everyone to Enovix Corporation's Second Quarter 2026 Financial Results Conference Call. Joining me today are President and Chief Executive Officer, Dr. Raj Talluri; and Chief Financial Officer, Ryan Benton. Raj and Ryan will be speaking to the slide presentation displayed on today's webcast. which will also be posted along with our press release on our Investor Relations website at ir.enovix.com. They will provide prepared remarks, and we will then open the call for questions.

Before we begin, please note that today's call contains forward-looking statements that are subject to risks and uncertainties. These statements are based on current expectations and may differ materially from actual future results due to a number of factors. For a discussion of these risks, please refer to the disclosures in today's press release and our filings with the Securities and Exchange Commission. You can also find these materials on our IR website.

All statements made on this call are as of today, August 12, 2026, and we undertake no obligation to update them, except as required by law. During the call, we may also reference non-GAAP financial measures. Reconciliations to the most directly comparable GAAP measures are included in the materials posted on our IR website.

And with that, I will turn the call over to Raj.

Raj Talluri

Good afternoon, everyone, and thank you for joining us. The second quarter showed momentum across all 3 of our primary target markets.

In smartphones, we reached an important qualification milestone. Our lead customer confirmed that our cells passed more than 1,000 cycles on the 0.2C discharge cycle test. We have one final cycle life test, and it's already well underway. We expect to finish remaining testing by the end of 2026 and with the customer acceptance and smartphone field testing to follow.

In smart eyewear, the production ramp of our lead customer has begun. We shipped approximately 2,100 batteries to a Tier 1 customer, recorded our first smart eyewear product revenue from this customer and expect to deliver roughly 9x that volume in the third quarter.

In drones, defense and industrial, we grew the pipeline by 41% from the first quarter. Drones led the way, we advanced product development and the drone pipeline alone exceeds over $100 million [ in Enovix ] now. Revenue for the quarter came in at the high end of our guidance, and we expect continued sequential growth in the third quarter. What is increasingly clear is that Enovix has multiple paths for growth that reinforce one another. Our 100% silicon anode AI platform is progressing towards smartphone deployment, while smart eyewear has entered commercial production. That sequencing is by design.

From the start, my strategy has been to go after the hardest market first, smartphones because meeting the most demanding energy density requirements forces us to build the best product. Everything we prove at the smartphone level then flows naturally into adjacent categories. Smart eyewear is the clearest example where the same platform is now shipping commercially.

Meanwhile, our MX platform takes that same silicon know-how and manufacturing discipline and aims it at defense, a big, fast-growing high-value market. Our established South Korea operations are serving defense and industrial customers today and their extensive experience with drones in particular, is opening substantially larger opportunities.

Briefly on the 2 platforms. The AI platform uses our proprietary 100% silicon anode architecture for space-constrained applications where the volumetric energy density and cycle life are most critical, while the MX platform blends silicon with graphite for greater gravimetric energy density and high-power performance manufactured at our proven facility in South Korea. These are not isolated businesses. They are mutually reinforcing.

And we're seeing this convergence translate into new areas for growth today. We are working on silicon blended opportunities beyond our traditional drone, defense and industrial markets.

Drilling down a bit further, First, on smartphones, we moved materially closer to completing qualification with our lead customer. The customer has now confirmed the cells passed more than 1,000 cycles under the 0.2C discharge cycle test. This is the same test our internal testing indicated when we shared it with you in February. The customer's own data has now borne it out. Fundamentally, this is a customer-confirmed evidence that the silicon anode batteries can perform at high levels.

Remaining work to be done is an accelerated cycle life test built around a hybrid protocol we defined in close collaboration with our customer to replace the traditional 0.7C testing approach for legacy graphite batteries. Testing is now live across several combinations of charge and discharge conditions as well with an enhanced cell design. And the same progression is underway. The enhanced cells are now showing stronger capacity retention and internal work and the data is now with our lead customers' hands for evaluation along with multiple variants of the hybrid protocol. We anticipate completing this final test in 2026.

Our second smartphone OEM is also moving towards a similar qualification framework, and we expect to begin sample deliveries in the fourth quarter. As we look towards 2027, we see the pattern repeating and expanding, our lead customer moving into commercial introduction with our second OEM advancing through qualification and additional leading OEMs with whom we are in active dialogue entering the qualification pipeline behind them. We pioneered the qualification testing pathway for silicon batteries and smartphones, so every customer after the first gets a faster, clearer path to execution and deployment.

I'm especially proud of our progress in smart eyewear, which has now moved from initial production into early commercial revenue with a Tier 1 customer. Recently, we completed a key international safety certification of our cells and battery packs as well as a full suite of customer reliability test. We shipped approximately 2,100 AI1 batteries in Q2 and recognized our first smart eyewear product revenue.

We have now delivery orders in hand for approximately 19,000 packs, which we are planning to deliver in the third quarter, a roughly ninefold increase from Q2. Those delivery orders are part of the customer's 50,000 unit pack order we expect to ship the remaining balance in the fourth quarter. Beyond 2026, we expect shipment volumes to grow as our customers' downstream deployments expand.

Turning to our defense sector. I'm proud of the team's execution from initial product launch in the first quarter to a substantial increase in our drone pipeline in the second quarter to customer sampling beginning now in the third quarter. The pipeline for products manufactured in South Korea increased 41% to approximately $183 million from the $130 million at the end of first quarter. As a reminder, this figure represents the estimated peak annual production value, the lifetime opportunities often many multiples more.

More than half of the growth came from growing opportunities, which now exceed $100 million on their own. Let me walk through what's inside that number. Because the funnel you see on the slide, more than $40 million of this pipeline is at stages where customers are actively evaluating and testing ourselves or designing them into products. And the breadth is striking, including some of the most recognized names in defense technology and consumer electronics.

We also introduced MX-1 to a broader set of customers and industry events in the United States and Europe. At approximately 360 watt hours per kilogram while supporting high continuous and pulse discharge, MXB01 (sic) [ MX1-B01 ] is designed to improve mission execution, flight time, range and payload capability. We've already ordered additional production equipment for the MX1-B01, and we expect it to be operational by mid-2027 with initial commercial shipments and revenue expected to follow as that capacity comes online and customer programs complete qualification.

This pipeline growth is also a commercial execution story. For the past 2 quarters, we've been deliberately building out our commercial organization, adding application engineers and product management talent, and we then brought in Steve Bakos, a seasoned sales veteran with more than 35 years in the global semiconductor industry with most recently running large global accounts at Infineon to lead our sales and application engineering teams under Samira Naragh, our Chief Business Officer. We are seeing the early results in that funnel.

Our South Korea operation is a meaningful advantage in pursuing these programs. It combines an established history serving defense customers with in-house manufacturing, quality and supply chain capabilities in a TAA-designated country. Our South Korea supply is TAA compliant today and ready for the expected mid-2027 capacity ramp, and we expect NDAA compliance across multiple product SKUs.

In July, our drone battery completed UN 38.3 transportation testing, creating an important step for commercial shipment, and we are commencing sampling with numerous customers in the third quarter. The next phase of Korea capacity is expected to come online in mid-2027, a very capital-efficient expansion, utilizing existing land and buildings we own and using readily available equipment.

The economics are attractive as well. ASPs are healthy. And because we own our own manufacturing, we believe the scaled volume can support solid margins. Beyond the current product, MX-2 remains targeted for 2027 with the goal of reaching 400 watt hours per kilogram.

Let me now come back to the AI technology platform. We produced the first AI-2 engineering samples in the first quarter. AI-2 is expected to provide approximately 20% higher volumetric energy density than AI-1 by combining thinner materials, better packaging efficiency and higher cathode voltage through our EX-3M technology node. We sampled cells to 1 Tier 1 smart eyewear customer in Q2. Many of the same EX-3M innovations are expected to carry into the future smartphone batteries and support another meaningful step forward in performance in that area as well.

I want to give you some insight into how our pace of innovation is also accelerating as it is something I'm particularly focused on. In batteries, the grading factor on development speed is cycle life testing. A full cycle life test has historically taken 4 to 5 months that sets the tempo of learning in the entire industry. We are developing AI models that can predict cycle life outcomes much earlier in the cycle life test than has historically been the case.

Our models for eyewear cells are getting close, and we're making very good progress on smartphone cell modeling as well. To be clear, customer qualification will always be the physical test, but this is about how fast we can learn and iterate internally. If we get this night, every design generation ramps faster and that speed itself becomes a durable competitive advantage.

Turning to manufacturing. The second quarter showed continued improvements across Fab2 with particularly strong results through most of the smart eyewear production flow. In fact, our smart eyewear cell output came in well ahead of our internal plan for the quarter, and our integral yield, the cumulative yield across the entire production line has now improved for 3 consecutive quarters.

Outside Zone 1, all but one process step operate at yields of at least 95% with individual steps as high as 99.6%. Zone 1 dicing remains our primary throughput bottleneck and a top focus, but the yield has improved to approximately 84% from 80% in the first quarter. Zone 1 has been a suborn constraint for a long time. This is exactly why we changed the approach rather than simply tuning it.

The hybrid dicing configuration uses laser and mechanical processes where each is most effective and is designed to lift Zone 1 throughput to multiples of today's rate. The step change we need to support the production volumes we are planning for 2027. Several of the key mechanical dicing steps are expected to come online around the year-end.

Supporting all of this execution is our growing team in India. A team in addition to conducting advanced research directly supports manufacturing execution at both Malaysia and South Korea factories.

Finally, I want to spend a moment on leadership because I'm thrilled to have Michael Vyvoda on board as the Chief Operating Officer. Michael brings decades of operations experience, including at Apple. He has a full scope mandate across manufacturing, supply chain, quality and customer delivery. His immediate priorities are increasing smart eyewear output, preparing manufacturing for smartphone field test builds and driving the cost, yield and delivery output improvements underway. Adding Michael gives me even more confidence that we have the right team for the next phase of scale.

With that, I will turn the call over to Ryan to review our financial results and outlook.

Ryan Benton

Thanks, Raj. We delivered another quarter of revenue growth and positive gross profit. We came in better than our operating loss guidance, and we ended the quarter with over $550 million in cash on the balance sheet, all while continuing to invest in the customer programs and manufacturing work that support the next phase of commercialization.

Second quarter revenue was $9 million, up 21% year-over-year and 19% sequentially at the high end of our guidance, our fifth consecutive quarter of year-over-year revenue growth. Defense shipments from South Korea remained the largest contributor, while smart eyewear generated its first product revenue, modest in amount, but an early proof point of contribution from AI-powered wearable devices.

GAAP gross profit was $1.3 million and non-GAAP gross profit was $1.8 million, representing GAAP and non-GAAP gross margins of 14.4% and 19.9%, respectively. The year-over-year decline in quarterly margin primarily reflected the mix of battery products sold through our South Korea operation rather than a change in underlying execution. Even with that change in mix, this was our seventh consecutive quarter of positive gross profit on both a GAAP and non-GAAP basis.

And the first half non-GAAP gross margin was up year-over-year to 22.8% from 21.3%. Non-GAAP operating expenses were $30.6 million compared with $28.8 million a year ago. The increase reflects continued spending on smartphone qualification, product development and manufacturing readiness, including support for the smart eyewear ramp. Non-GAAP loss from operations came in at $28.8 million, better than our guidance range of a loss of $29 million to $32 million.

Adjusted EBITDA was negative $18.9 million compared with negative $20.1 million in the second quarter of 2025. And non-GAAP net loss per share was $0.13 at the favorable end of our guidance range of a loss of $0.13 to $0.17 and unchanged year-over-year.

Turning to cash flow. Net cash used in operating activities was $21.8 million, down from $25.9 million in the second quarter of 2025. And free cash flow was an outflow of $31.4 million versus $33.8 million a year ago, both better year-over-year despite higher capital expenditures supporting our manufacturing scale-up. The operating improvement primarily reflected favorable working capital changes.

Capital expenditures were $9.6 million, principally supporting manufacturing readiness and capacity expansion. We ended the quarter with approximately $552.1 million in cash, cash equivalents and marketable securities, including restricted cash. That liquidity allows us to fund the qualification and commercialization milestones already underway while preserving flexibility for selective strategic investments.

We did not repurchase any shares during the quarter. Our capital deployment priorities remain unchanged, product qualification completion, disciplined manufacturing investment and commercial execution.

For the third quarter, we expect revenue between $9 million and $10 million, up approximately 13% to 25% year-over-year. The range assumes continued defense and industrial shipments from South Korea and a significant sequential increase in smart eyewear deliveries. We expect non-GAAP loss from operations between $29 million and $32 million and non-GAAP net loss per share between $0.13 and $0.17. We expect capital expenditures between $8 million and $12 million, primarily for Fab2 initiatives and South Korea capacity expansion Raj discussed.

As always, quarter-to-quarter revenue and gross margin can vary based on product mix, customer delivery timing and the pace of qualification and commercial program ramps. For the third quarter specifically, 2 factors will shape gross margin, product mix in our South Korea business and the early cost of the smart eyewear ramp before volumes reach scale. We will continue to manage spending with discipline and align our investments with measurable customer product and manufacturing milestones.

And with that, let me turn the call back over to Raj for some closing thoughts before we open the call up for questions.

Raj Talluri

Thank you, Ryan. This quarter, all 3 of our markets moved forward at the same time. Smartphone silicon batteries passed a critical milestone; smart eyewear entered its revenue generation stage; and our drone and defense pipeline is growing rapidly. The milestones to watch from here are just as clear, completing the final accelerated smartphone qualification test, initial sample deliveries to a second smartphone OEM, the smart eyewear ramp, converting drone and defense opportunities into design wins and continued improvement in manufacturing, throughput and cost.

With that, operator, we're ready to take questions.

Operator

[Operator Instructions] Our first question will come from Colin Rusch with Oppenheimer.

分析师问答

Colin Rusch

I just want to get a sense of volumes as you start to ramp the eyewear business line, how do we want to think about total volumes to get to that optimal margin level? And how many quarters do you think it will take to get there?

Raj Talluri

Yes. Thank you, Colin. I can take a shot at it, and then Ryan will add some more commentary on margins and so on. Firstly, we're actually very pleased with the progress in manufacturing on the small cells, right, going from a few thousand packs to now roughly 19,000 next quarter and on the way to fulfilling a 50,000 pack order. This shows our confidence in manufacturing of the cell on our technology platform in our Penang fab. And I'm really proud of what the team has done there.

Look, the market itself, as I said the last quarter, the market is multiple millions of units and expected to grow year after year. We are sampling now to different customers who are in like various stages of building the products. Exactly how much we'll ship will depend upon how successful their products are and what share we win. But I can say this, the feedback we've got from the customers is the cell has been very strong, mainly driven by the energy density we provide in the small form factor. And it's translating into much longer battery life, particularly with AI running on these glasses.

So maybe, Ryan, do you want to take on the margins and?

Ryan Benton

Yes. No, that's fair. I mean, look, not to be repetitive, but we shipped 2,100 units in Q. The absolute revenue from that is nominal, of course. On pricing, specifically, I'm not going to quote a number there, particularly because we're dealing with one single important customer. And so I can't really go into those economics. But what matters for us really, I consider most important is that we build scale is that we get into a position of incumbency into the growing market. We want to be the default better of choice for smart eyewear, specifically kind of finishing off on margins.

Of course, even at the 50,000 unit level for the year, that can absorb the overhead burden that it will face as those costs start to move in geography from operating expenses up above the gross margin line. So we expect it to be negative margin for the balance of the year. It really as we ramp ultimately as we get to some version of scale, and I'm not going to quote a specific number of what the breakeven point is. Raj talked about the market growing as the customer adoption, the pull-through starts to happen, we do expect to have healthy gross margins.

Colin Rusch

Okay. Excellent. And then in the drone market, obviously, there's a lot of different applications. And so I want to get a sense of how you're sitting within that opportunity and which applications you're competing well on? And what are the key drivers -- which products are really driving some of that progress that you're making in that end market?

Raj Talluri

Yes. On the drone market, we are seeing tremendous amount of interest from many customers. One of the main reasons is we now have an extremely competitive cell with -- there the main metric is watt hours per kilogram. And that cell actually is made fully in our own factories. And it's a TAA-compliant country in South Korea, and we expect it to get to NDAA compliant path across multiple SKUs in 2027.

And this factory, you might remember, Colin, has more than a decade of production history on this site, so into military applications. The main markets we are getting into there are markets where they want a few hundred cycles, for example, of flying time and also safety, public safety, interceptor drones, ISR and markets like that. These are the markets that I feel like have a good margin profile and a lot of demand.

And our expectation is that the demand is actually going to outstrip the supply here very quickly with many, many customers wanting that. And we are super excited that we have this -- our own factory that we're able to make it in where the margin profile will be really good. And we are adding more capacity there. And as I mentioned in the prepared remarks, we expect that capacity to come online in mid-'27.

Ryan Benton

And if I could chime in as well. I mean, as we go through the names of the pipeline and look at it, these are some of the best companies in the world. So some really exciting opportunities. Of course, drones is a big portion and the majority of that funnel that's building right now, we expect that to be strong for many years to come. But then there's other technologies that we think will build right on top of this. So I think robotics is another example that we look a few years out, we think it's going to be a big market as well.

Operator

Your next question will come from Ruplu Bhattacharya with Bank of America.

Ruplu Bhattacharya

Can you guys hear me now?

Ryan Benton

Yes, sir.

Ruplu Bhattacharya

I had 2 questions. Raj, now that both the lead smartphone customer as well as the second smartphone OEM, they've moved to a silicon-specific qualification framework. How standardized do you think this will become in the industry? And what steps are you taking to enable that?

And does the second OEM now have essentially the same qualification path as the lead customer? And when do you expect POs from the first customer as well as the second customer? And I have a follow-up.

Raj Talluri

Yes. So Ruplu good question. Yes, I think what has happened is over the -- what we've done over the last couple of quarters, we've really been able to convince the customers that when you move from graphite anodes to 100% silicon anodes, some of the legacy tests are not very representative of how the battery will actually perform in real life, which is, I think, a huge step forward that we've been able to accomplish.

And as I mentioned, in February, we talked about this 0.2C cycle life test where we're internally seeing that we should be able to go past 1,000 cycles. Now the customer has confirmed that on their own test. So we are very happy about that.

And the one test that's left is really the accelerated cycle life test where if you just do at a normal cycle life, 0.2C, it just takes a long time. So people really need an accelerated test. We've now worked with the customers to come up with a hybrid protocol, which is some combination of the different rates of discharge. And that's what the customers are running -- my lead customer is running now. There's 2 or 3 different protocols that we work with them on. And our expectation is one of them will meet the requirements that they have in the fourth quarter.

Then we expect to get to a small build of -- they'll put the battery inside their phone to see how it performs. And then we start getting into volume production in '27 and so on. We do continue to talk to our other customers, too, and then we talk to them about -- we have a good engineering relationship where we talk about these protocols. I do expect in time that the market will adapt and change to these kind of protocols because silicon behaves differently than graphite in accelerated tests.

Ruplu Bhattacharya

Okay. As a follow-up, can I ask, you've made many manufacturing improvements in Fab2. What is the manufacturing capacity now as it stands today of HVM-1 as well as the agility line? Can you give us a sense for like how much max units of smartphones and eyewear that each of these lines can support?

Raj Talluri

Yes. I mean, look, like I said, when we first said what the lines were, we have continued to keep that. 1,350 UPH was what the nameplate capacity of the line was. We haven't really staffed to all of that because we are managing that through the qualification time line with the customers so that when the demand is there, we are there. And again, it's not really a question of how much capacity we have. It's more a question of pacing the line with the right number of people and working on the yields and working on the ramp in line with as the customer qualifications are going. So that's kind of where we are on that.

Operator

Your next question will come from Mark Shooter with William Blair.

Mark Shooter

You can hear me, right?

Ryan Benton

Yes, sir.

Mark Shooter

Awesome. Raj, congrats on passing the 1,000 cycles. That's a big accomplishment. I do think that we thought that this would be the last milestone stage gate, though. So can you walk us through what the last test, what are we trying to prove there? I mean I know it's a hybrid approach, but is there something that your lead smartphone customer saw that required a little bit more digging in? Any color there would be helpful.

Raj Talluri

The last test basically is what I call accelerated test, which basically means that they would like to discharge at a faster rate, as I mentioned, but not just all the way at the high rate, but some combination of a faster rate and some combination of a slower rate, which is kind of more representative of what a phone might actually do in the real world. And we are working with them on the protocol that actually will get that done, and they have a number that we need to meet to get to that. And that's what we are working on.

The next stage after that is to actually do the test inside a phone, right? These are bench tests on the battery on the table. So that's the next step to it. Like I said, the main gating item right now is an accelerated cycle life test so they can complete the test in time. And we have multiple protocols that we are working jointly with the customer, and they're all underway. And by fourth quarter, we expect to see some good results.

Ryan Benton

And Mark, at the risk of being repetitive again, it's like go back to the February print, and we showed in the presentation and talked about how we passed the 0.2C test internally. We're waiting on the customer to run that test themselves. And that's what we've put in the headline here today, and we're very proud of that we've got that independent verification from the customer.

What we also talked about in the February print and follow-up in the May print was the 0.7C test, and that's the traditional graphite test that we need to define a proxy or a substitute. And now we've got the framework in place we mentioned last quarter. And now we have cells running under a handful of separate variant protocols, and we expect one or more of those to pass by the end of the year.

Mark Shooter

Okay. Got it. Switching over to the drone opportunity, which is increasing here with a $100 million pipeline. It's a big number, and it's an exciting business. The -- what I'm interested in is, can you walk us through maybe some of the time frame of what that engagement looks like with the customer.

So you have $5 million in wins already. But can you walk us through how long it may take to move somebody from a pipeline opportunity to down subsequent steps? Is it 9 to 18 months? And when do you see some of a decent conversion from that pipeline opportunity into backlog?

Raj Talluri

Yes. Like as I mentioned, right, I think we have a pipeline that has grown significantly. The $5 million is only just design win, awarded one. The better number is like there's $40 million already in active testing and design win for the customers to our cells right now.

Defense qualification cycles are shorter than smartphones because just of the need -- of what they need to get to production faster. And the revenue and scale aligns with our Korea capacity coming online in mid-2027. So we start to expect to see ramp of this -- some of this pipeline in the mid-'27. And again, this capital expansion is very capital efficient because it's on the land and buildings we already own.

Ryan Benton

Yes. And if I could comment as well because I said to those same pipeline reviews, it's -- there's some splits within that group as well, right? So the drone companies themselves, it's -- they're all trying to move really fast, in my opinion, in the sort of 6 to 9 months and we're sampling those. And quite frankly, within our internal teams and cell teams, everyone is fighting over samples in terms of how we prioritize.

And then there's a separate split of the defense primes. And as we start to focus on supporting those type of activities, those tend to be a little bit longer runway. So 18 months is, I think it's not an unfair number to say what we think the average time is to get to production with some of those. But those represent some enormous opportunities. And so in both of these areas, it's really about us putting capacity in place and the equipment that we've ordered is in flight right now, pardon the pun, is just hopefully the beginning.

Operator

Your next question will come from Derek Soderberg with Cantor Fitzgerald.

Derek Soderberg

Just a clarifying question here first, Raj, you were talking about the testing earlier. Just wanted to confirm the second smartphone OEM is accepting the 0.2C testing standard? Or are they -- I think you had said there are sort of a few different options, but just wanted to confirm that second smartphone customer is sort of accepting the 0.2C.

Raj Talluri

Yes. I mean, look, we are focused on the first one first. And when that one gets to the right stage, we will sample the second one. But we have talked to all of them about silicon being different, and they all understand that. And I think some of them have launched some amount of silicon batteries already. So they do understand that they behave differently. So I think, my expectation is the whole market will move towards that in time.

Ryan Benton

And it's fair to say, each one should be easier and easier. We are not planning [ first to fail, ] yes.

Raj Talluri

Because once you do first time right, yes.

Derek Soderberg

Got it. That's helpful. And then just on the Zone 1 yields improved quite a bit since Q1. I was wondering if that was mostly the dicing configuration. And then just kind of a high-level question on yield. Where do you guys feel like you need to get before you can really ramp up production? Maybe a yield number that gives you the confidence to invest in additional capacity just with the assumption that demand is not the issue.

Ryan Benton

Maybe I'll take a first stab at that one. I think, Raj has talked for several quarters, and I've emphasized as well in terms of we're making steady progress, and we're doing things in kind of an orderly fashion in order to kind of meet our customer commitments. And so, I think there's just been -- it has been time to put a lot of steady progress and a lot of focus on just kind of grinding out disciplined yield wins. And so there's a lot of tactical things that go in to represent and be reflected in that 4 percentage point increase.

So there's mechanical changes, there's process changes, there's a lot of different things. And the team in Malaysia has done just a fantastic job grinding out those wins. And Michael, who has joined as COO, we think is a great addition and adds to the team, and he's over there in Malaysia this week and reporting good things. And I think he's going to just help with the momentum of progress.

In terms of yield -- to be able to start a ramp, well, I mean, we're starting the commercial ramp with Smartware. So we feel good about it. And we feel good that we're on a glide path, and we've got a path to the gross margins that we want to need. And I don't think the margins that we would target is as baseline margins in order to start a real high-volume ramp is going to be any different than any other typical manufacturing concern.

Raj Talluri

And it's just continuous improvement. Yes.

Ryan Benton

Fewer words, yes, that's better.

Operator

Your next question will come from Bill Peterson with JPMorgan.

William Peterson

Maybe picking up on that last topic. So you have the new COO, Michael. Do you expect that he'd be probably more focused on Malaysia, improving the areas that you just spoke to or Korea or somewhere in the supply chain? Just kind of any sort of tangible area where you think that you can get the most continuous improvement using your words.

Raj Talluri

Yes. I mean we don't -- I think we mentioned last time with KH, who came to us with a tremendous experience from our Routejade acquisition is now responsible for the manufacturing of both the factories. So Michael will be responsible for both the factories in addition to advanced manufacturing machines that we need to build, in addition to supply chain, getting the right materials in place. So the entire operational side will be under him.

And I said in the prepared remarks, our Malaysia factory is benefiting quite a bit from our learning in Korea because we've done that battery manufacturing know-how. And with KH being there, the cross-pollination is happening tremendously. And also, our Korea factory is benefiting tremendously from our silicon knowledge and how that's actually helping us make a very competitive drone cell.

And our India team is actually helping both of those. We have tremendous R&D team in India. So it's kind of a holistic set of teams that are all working together, and we don't separate that much as Malaysia and Korea as much as battery know-how across both those.

William Peterson

The next one, I guess, maybe probably for Ryan, but the gross margin took a step down in the second quarter. You called mix primarily sold from South Korea. Can you provide some more color on that? And I guess it sounds like you didn't really have any impact yet from the small volumes of the eyewear, but it sounds like based on your expectations that might be a bigger impact. So how should we think about gross margin trajectory based off the prior comments around eyewear and maybe other mix ramifications from South Korea?

Ryan Benton

Yes, good question. I mean, look, first, I'd caution against reading 1 quarter as a trend. I think more appropriately, if you look at the first half, on a first half basis, non-GAAP gross margin was 23%, I think, versus 21% in the prior year. So year-to-date, actually up 2 points. Q2 specifically, again, reflects the product mix and principally in the Korea base business, which can be lumpy. The -- some SKUs carry better margins than others. It's just the case.

Looking forward, we don't -- obviously, we don't guide gross margins. But the shape to understand is that the base business is there. Last year, if I recall, '25, Q4 ended up being a stronger margin quarter than Q3. But there's no doubt, smart eyewear, as we ramp, it won't be big numbers, but it will drag -- it will be a drag on margins as we move into that ramp as the overhead gets moved into gross margins. Again, from an operating income, from a cash flow basis, it's largely geography moving. And really, again, as we scale to '27 and beyond, we expect that to rationalize as we scale.

Operator

Your next question will come from Ananda Baruah with Loop Capital.

Ananda Baruah

I apologize if this has been asked already. Raj, memory availability, does that have any impact just because of the prolonged tightness on any of the -- like the SKU qualifications on smartphone or PCs that you guys are targeting? And then I have a quick follow-up.

Raj Talluri

Not much for us yet. We're not in high volume. So we are really in a qualification stage. So I think we don't see too much impact right now. Yes, the customers are worried about it, but not impacting us.

Ananda Baruah

And is there a -- there is a component of what's going on where low-end phones, low-end SKUs are unable, they are just exiting the market to some degree. Would that impact you at all? Where exactly would you consider yourselves to be positioned inside of sort of that heat map, that SKU heat map?

Raj Talluri

I mean, really, we're in technology qualification stage and people really want to use us in the leadership products because that's where they see most advantage of our technology and differentiation, and that's where we are being qualified. But I do expect that to waterfall down. And like I said, at this point, the SKU mix is not impacting us that much because we are working in the qualification stage. That will come in time, but not right now.

Ananda Baruah

Cool. And if you -- sort of just -- if you -- and this is for our benefit, but if you thought about what your revenue mix is 3 years out, 5 years out like that, different product segments, how -- what do you think is a useful way for us to envision what the rev contribution is to the company? Maybe like sort of anecdotal percentage, percent -- percent this, percent that, not necessarily a rev outlook.

Ryan Benton

And maybe I'll try and approach. Yes, of course, we can't necessarily quote it -- we can't quote a number or specific. I will say, it's going to be, I think, a pretty good horse race. A year ago, if I think you asked that question, we were talking principally about smartphone would be the obvious answer. I think the way that the drone market and the defense market has developed and the progress that we've made with product development, customer qualifications and building the team, that has an opportunity to be a really big business really fast.

Raj Talluri

Yes. And again, smart eyewear could take off and become huge, and that could be good, too. So it's hard to call the mix, but I think all 3 markets for us are attractive and where our technology provides clear benefit to our customers. So that's probably the best way I'll say it. It's exciting times because everybody wants better batteries.

Operator

[Operator Instructions] Your next question will come from Jeff Osborne with TD Cowen.

Jeffrey Osborne

Just a quick one, Ryan. Could you update us on where annual production capacity is at the South Korea facility now? And then Raj mentioned that you would be expanding, and that would be up and running, I think, by the middle of '27. So I was just curious, where is it now annual revenue capacity? Where are you headed? And then how much will it cost?

Ryan Benton

Okay. I think I got all the subpart. Okay. Thanks for the question. In terms of capacity there, I mean, we're not operating at full capacity. So there's some headroom there. There is some complexity a layer down because there's different sets of equipment. So each different equipment line has different -- some are at capacity, some are not. That's the reality of the current building.

In terms of the capacity for the drone business, we have minimal right now. We have placed orders for new equipment that gets us what we think is significant material capacity coming on in the summer. And that equipment is going to be tailored and customized to -- standard equipment, but configured for our specific product SKUs.

Jeffrey Osborne

I guess, so are you preparing for this to be like a $100 million business 18 months from now? Is there any flavor or sense of...

Ryan Benton

Yes, I can answer. The capacity of that equipment, I guess I can quote that number is roughly 1 million units. And it's all embedded in the CapEx forecast that we're starting to make payments on. And you see, I think it's $8 million to $12 million that we guided. So it's -- as Raj cited, capital efficient. We have additional buildings and land there in South Korea, so we can add incremental capacity beyond that. So hopefully, this is just the first inning of that baseball game.

Raj Talluri

Yes. And our goal is to make sure that as Samira and her team converts the pipeline into opportunities into wins, we don't get capacity limited, right? So -- and we are staging it in that way. We are building the factory in that way so that incrementally, we can add capacity quickly. But up to 1 million units a year, as Ryan mentioned, we are already on track. So we can keep driving more on that as we see the design wins come in. We have the space and the building, so it's really not a problem.

Operator

There are no further questions at this time. With that, I'd like to turn the call over to Dr. Raj Talluri for closing remarks.

Raj Talluri

Yes. Thank you all. Really great quarter. We're happy with all the progress we've made and look forward to seeing you next quarter.