Prepared remarks
Greetings. Welcome to the Palladyne AI Corp. Second Quarter 2026 Earnings Call. Please note, this conference is being recorded. I will now turn the conference over to Brian Siegel, Senior Managing Director, Hayden IR. Thank you, Brian. You may begin.
Thank you, Nicole. Good morning, and welcome to Palladyne AI's Second Quarter 2026 Earnings Conference Call. Joining me on the call today are Ben Wolff, President and Chief Executive Officer; and Trevor Thatcher, Chief Financial Officer. Earlier this morning, Palladyne AI issued a press release announcing financial results for the second quarter ended June 30, 2026, along with the updated commentary regarding backlog and its reiterated 2026 revenue guidance. A copy of that release along with the accompanying financial tables is available on the Investor Relations section of Palladyne AI's website. Today's call will include prepared remarks from Ben and Trevor followed by a Q&A session. During today's call, management will make forward-looking statements within the meaning of the federal securities laws. These statements include, but are not limited to, statements regarding Palladyne's 2026 revenue guidance, expected backlog conversion, anticipated quarterly operating cash burn, product development milestones, commercialization timelines, defense program activity, potential customer adoption, market opportunities and future strategic positioning across aerospace, land and maritime domains. Forward-looking statements are based on current expectations, assumptions and beliefs and involve risks and uncertainties that could cause actual results to differ materially from those expressed or implied. These risks and uncertainties include, among others, Palladyne AI's ability to execute on development programs, convert backlog into revenue, scale production, manage operating expenses, integrate acquired businesses, secure additional contracts, maintain liquidity and navigate evolving commercial market conditions. These and other risk factors are described in detail in Palladyne AI's filings with the Securities and Exchange Commission, including its annual report on Form 10-K and subsequent filings. Palladyne undertakes no obligation to update any forward-looking statements, except as required by law. In addition, during the call, management will reference certain non-GAAP financial measures. In general, management will adjust for acquisition and other transaction-related expenses, stock-based compensation expense, noncash warrant income or expense that are mark-to-market quarterly based on changes in the company's stock price, expenses related to the change in contingent consideration liabilities associated with closed acquisitions and any tax impact these items may cause. A reconciliation of these non-GAAP measures to the most directly comparable GAAP measures is included in this morning's press release. With that, I'll turn the call over to Ben.
Thank you, Brian, and good morning, everyone. Thanks for joining us. Every quarter, we meet with our Board of Directors to review our prior quarter and year-to-date performance and to discuss, among other things, the business's prospects, opportunities and challenges in the near, mid and long term. Prior to that meeting, I meet with the various leaders across the organization for a comprehensive review of the same topics at a business unit and division level. Our most recent Board meeting was last week, and I have to say, the most recent round of quarterly meetings validated the growing optimism I developed over the quarter. I'm feeling more bullish about our prospects than I have since I returned to the company almost 2.5 years ago. We are seeing meaningful traction across all parts of the business. Since I returned, we've built the two things that I believed we needed in order to deliver on our vision for making the United States defense industrial base more competitive, more capable and more responsive. The first is a new kind of autonomy, a decentralized embodied collaborative AI architecture that enables machines to operate on their own, work together and adapt in real time without being preprogrammed or relying on a constant connection to the cloud or relying on human direction and intervention. What I'm describing is the definition of true collaborative autonomy. What most others mean when they use these words is really nothing more than preprogrammed action where humans have made all of the decisions for the machines well in advance. The second is the ability to make the machines that can put that autonomy to work, which requires world-class aerospace engineering, low-cost next-generation avionics, precision manufacturing capabilities and ultimately complete weapon systems, including access to some of the most capable battle-proven loitering munitions on the planet. Together, they will enable us and our customers to field affordable, attritable systems at scale that are designed from the ground up to leverage our decentralized embodied collaborative AI. These two objectives are closely aligned with our nation's priorities. The Department of Defense has made it abundantly clear that the United States needs both more intelligent autonomous systems and a stronger industrial base capable of designing, manufacturing and fielding them quickly and economically. We are focused on doing our part to answer that call. This quarter, we began to see every one of these pieces working together in a way that is showing up in revenue, customer activity and a number of opportunities in front of us. We still have much to do, but the business has meaningful momentum across the company. We generated record quarterly revenue of approximately $5.8 million, up 63% from $3.5 million in the first quarter. The growth was broad-based across our business units, and we expect continued growth through the second half as backlog converts to revenue and new contract awards and customer wins begin contributing. We ended the quarter with approximately $25 million of backlog, up from $17 million at the end of the first quarter, which means after taking into account our Q2 revenues, we booked roughly $13 million of new business during Q2. That backlog will not necessarily increase every quarter. The timing of the bookings, the duration of contract performance and the timing of revenue recognition can all cause backlog to ebb and flow. But the current backlog and the level of new contract awards and customer engagement give us good visibility into the second half. As a result, we are reaffirming our full year 2026 revenue guidance of $24 million to $27 million. We ended the quarter with approximately $44 million of cash on hand. Operating cash burn, including capital expenditures, was above our guided average quarterly rate primarily due to increased business development efforts, some nonrecurring capital expenditures and the expansion of our engineering team to support new business that will contribute to revenue in the second half of this year. We are also carrying infrastructure that our current revenue base has not fully utilized. This cash burn was offset by our prudent use of our ATM facility. We continue to expect operating cash burn of $32 million to $36 million for the full year, inclusive of CapEx, which means we expect operating cash burn to decline in the second half. The defense market is moving quickly towards affordable mass: large numbers of lower-cost autonomous systems that can be deployed and replaced without the economics of traditional weapons platforms. The Department of Defense is also pushing the industry to develop and field those systems faster than ever. Both trends are directly relevant to what we have built. For the most part, launching 1,000 drones today requires 1,000 soldiers. Other companies are attempting to solve this manpower challenge by automating multi-drone launches that follow a preprogrammed flight path. But that doesn't really solve the challenge because most of these drones can't react to what's happening around them in real time. That still requires a soldier. Preprogrammed flight, or what we call automated flight, is not a substitute for human intelligence or human direction, but true autonomous flight can be. That's what we do, and it is essential to understand the distinction and how that distinction translates to real-world operations. We put real reasoning and decision-making abilities directly on the drone and then we enable multiple drones to collaborate in a manner such that the knowledge of each individual drone can be aggregated and used across the entire swarm. This is what we mean when we refer to true collaborative autonomy. And we don't just do this on Palladyne drones. We enable UAVs from different manufacturers to collaborate with any drones that are also running our SwarmOS software. We call our approach Decentralized Embodied Collaborative Autonomy, or DECA for short. The intelligence operates on the machines at the edge so they can coordinate and adapt without relying on a continuous connection to the cloud or from a remote operator. We proved this capability in the field under real operating conditions during the Army's Ivy Mass exercise in Q2. Soldiers from the Army's Fourth Infantry Division used SwarmOS to command a mixed team of surveillance drones and our own Gremlin-X drone in a contested environment. One Army operator controlled the entire group of UAVs. In June, we announced that SwarmOS and Gremlin-X were awarded contracts by the Army under its Disruptive Applications Program. Following Ivy Mass, we were informed that certain contract options would be exercised for FY '27, and we were formally invited to additional exercises. That exercise and the future exercises are important steps for us because the software was and is being used by soldiers in an operational setting, not simply being demonstrated by our engineers in a controlled environment. This was not a demo in the traditional sense; rather, it was a real-world exercise. A few moments ago, we issued a press release about our success at PC-C6, which was even larger and more complex than Ivy Mass. Turning to our business operations, I will start with our components business. BRAIN is our low-cost flight computer for autonomous air systems. It is built into our own Gremlin-X mini bomber drone, and it is being integrated on systems designed by us for our defense prime customers. During the quarter, we booked a follow-on order from a defense prime that was five times larger than its prior order. The order covers more than 1,000 BRAIN FC1 units with an aggregate value of approximately $2.3 million. These units deliver the guidance, navigation and control for a new low-cost counter-UAS interceptor. As these systems gain traction with customers, we expect demand for BRAIN to grow with it, and we are expanding our domestic production capacity accordingly. BRAIN began in building our own platforms like Gremlin-X and SwarmStrike, and while designing third-party systems we have developed a range of other high-value subsystems and components, all made here in the U.S. We are now exploring the business case for supplying some of these components to other drone and missile manufacturers as well, but we will only do so where we have a unique competitive advantage and can deliver differentiated capability or performance or a cost advantage. We have no interest in being in a highly competitive commoditized or low-margin business, although there can be no doubt that as a nation, we need more UAV and missile components made here to reduce reliance on foreign sources. Our precision manufacturing business accounts for the balance of our components revenue. It machines and fabricates hardware for defense programs and industrial customers. This capability is strategically important because it allows us to produce components that are often in short supply and that impair our country's capacity to produce the volume of weapons that the Department of Defense is looking for. We also expect them to support the work required to bring IAI's loitering munitions into production in the United States when and as we develop customer demand. Our engineering services for UAVs, missiles, loitering munitions and spacecraft generate real revenues today and give us a seat at the table to educate and inform customers about the component and software solutions we offer that could give these customers and their products a competitive edge. During the quarter, we added engineers and moved into a larger facility to accommodate the increasing demand and utilization rates we are experiencing. Specifically, the additional capacity allows us to support existing and new third-party programs, continue developing our own products and manufacture and assemble BRAIN and Gremlin-X without one activity interrupting the other. SwarmOS also continues to mature and is increasingly being integrated with autonomous systems from other OEMs. An operator defines the mission, but the software allows the machines to respond to what is actually happening around them, coordinate with one another and adjust in real time. It is not tied to a single airframe or manufacturer, which is one reason the Army has asked us to deploy our software across multiple drone platforms. We are also extending the technology to integrate sensors from other domains, including sensors in space under the recently signed $4.2 million HANGTIME contract with the U.S. Air Force. Today, revenue relating to our DECA AI solutions still comes primarily from development, integration and trial contracts with the Department of Defense, but customer engagement is increasing and the software remains the foundation of our aerospace and defense strategy. Our Autonomous Systems portfolio combines that software with physical platforms. Gremlin-X is a reusable mini bomber designed to deliver a kinetic effect for less than $1,000 per effect. We have prioritized development of this platform, which has advanced from TRL 6 to TRL 7 in just the past quarter. We're also evaluating a modular version that could support missions beyond strike, including potentially ISR and counter-UAS missions, by making the payload modular rather than having to field different airframes for different missions. SwarmStrike is our low-cost mini cruise missile designed to operate in coordinated groups using our IntelliSwarm product, which combines SwarmOS and BRAIN. With our ALRRM contract with the U.S. Navy, we are in the early stages of developing a low-cost air-launched near-hypersonic missile for use with the F-35. During the quarter, we were one of only 14 companies invited to the Air Force's Relentless Wolfpack Industry Day, which focused on networked, autonomous weapon salvos. We had proposed SwarmStrike as an answer to the challenge, and we were the only small company invited to participate. This does not mean we will land a contract, but it shows the Department of Defense recognizes our innovation and capabilities. Separately and importantly, a major Tier 1 defense prime included SwarmOS in its own Relentless Wolfpack proposal. Again, we don't know who will be down selected but this is great validation from a large industry partner. Major defense primes don't make these kinds of decisions lightly. I will now turn to our partnership with Israel Aerospace Industries, or IAI. In June, we announced that Palladyne had entered into a partnership with Israel's largest defense prime. As part of that partnership, we secured the right to Americanize, manufacture and integrate IAI's HAROP, HARPY and Mini HARPY loitering munitions for the U.S. Department of Defense. These are large combat-proven systems, not small FPV drones. IAI pioneered the loitering munition category more than 40 years ago and today has a backlog across its business of more than $30 billion. HARPY and HAROP, which have been sold in numerous countries around the globe, are 10-foot wingspan unmanned aircraft that are designed for long-range missions, extended loitering and the suppression and destruction of enemy air defenses. These are exactly the kind of systems that could potentially mitigate some of the aircraft losses our country has recently experienced in the Middle East. HARPY is designed to search for and attack radar emitters. I am not aware of any other UAV of this size or class that has anti-radiation capabilities. HAROP has an electro-optical seeker allowing the HAROP to identify and engage a target with precision. Mini HARPY combines anti-radiation and electro-optical capabilities in a smaller system with shorter range and at a lower price point. These products have been used in combat and address missions that have become increasingly important as mobile air defense systems become more common and present our military with real targeting challenges. We believe the U.S. military has a meaningful capability gap in this area. As a nation, we typically counter these defenses by launching missiles or by using manned aircraft. I have been told, but have no way of independently confirming, that the current cost of striking one of these mobile or intermittent air defense systems can range from $10 million to $40 million. Our IAI systems can present an alternative at a small fraction of that price without risking human life or the cost of a manned aircraft. We believe that a successful U.S. program based on these IAI systems could ultimately amount to hundreds of millions and potentially even billions of dollars of revenue over time. IAI selected Palladyne based in large part on our engineering capabilities, our U.S. manufacturing capacity, our team and our autonomy software. Once there is a clear government demand signal and we begin to work to stand up domestic production, we believe initial units could be available in about 18 months. We also intend to integrate SwarmOS where appropriate so these systems can coordinate their actions as part of a larger mission set rather than operate only as individual aircraft. That combination takes the mission capability to a level unlike anything else on the battlefield today. One more proof point worth mentioning is the caliber of people joining our team. This quarter, we added senior military advisers to our Defense Advisory Board, including retired Lieutenant General Sean Gainey, who ran U.S. Army Space and Missile Defense Command, led the operational command behind the Golden Dome and founded the U.S. military's joint counter-UAS office. People with that kind of record don't lend their names to companies they don't believe in. He joined us precisely because he truly believes in our mission, our capabilities and our opportunity to make a real difference for our nation's warfighters. On the commercial and industrial side, we have sharpened our strategy with Palladyne IQ in two ways. First, we are focused on applications where autonomy addresses work that is difficult, dangerous, repetitive and hard to staff, like surface preparation and corrosion treatment typically done by hand. A good example is our robotic work at an Air Force Logistics Center, which we expanded this quarter with an additional $2.9 million contract award, bringing the total for this program to more than $10.6 million to date. Second, we are shifting to a partner-led go-to-market strategy for IQ and Pilot rather than trying to sell and deploy every solution ourselves, and we are seeing encouraging interest from ecosystem partners. We are not ready to announce more today, but we expect to have more to share on both products over the next several quarters. I will finish with the opportunity pipeline. As of today, we have submitted proposals for more than 20 Department of Defense funding programs with an aggregate initial potential value of several hundred million dollars. The timing and probability vary significantly by program. Some are relatively near term; others are larger opportunities with longer timelines. Of course, we will not win all of them, and we may not even win any of them. But the important point is that the number and size of the programs we are pursuing are materially greater than they were six months ago. We are increasingly getting shots on goal as the Department of Defense's requirements continue to evolve and begin to incorporate the types of capabilities we have been developing for years. To continue the analogy, we have been skating in front of the puck for a while now, and I believe we are just on the cusp of that strategy and focus paying off. We have a lot of work ahead of us and much of the opportunity I discussed still depends on successful execution and customer funding decisions. But the quarter showed progress in the areas that matter: revenue, contractually committed backlog, proving out our autonomy software with customers in the field, repeat component orders and a much larger opportunity pipeline. The pieces we assembled over the last two years are beginning to reinforce one another. That is why I'm increasingly confident that we are in the right place at the right time. With that, I will turn the call over to Trevor to walk through the financials in more detail. Trevor?
Thanks, Ben. I'll focus on our second quarter results, our liquidity position and our capital outlook. Revenue for the second quarter was $5.8 million, up 470% from $1 million a year ago and up 63% sequentially from the first quarter. Product revenue, which today comes primarily from our precision manufacturing business and our BRAIN avionics product, was $2.7 million. Engineering services revenue was $1.9 million, and AI product development contract revenue was $1.2 million, which accelerated near the end of the second quarter after the government shutdown suppressed revenue during the first quarter. Cost of revenue was $4.1 million and consolidated gross margin was approximately 29%. Similar to the first quarter, our margin reflects manufacturing utilization running well below full capacity and includes setup costs and investments related to new program wins that have not yet been covered by revenue. As these programs come online, we expect that these efforts will turn into a tailwind as incremental volume runs through capacity we're already carrying. Research and development expense was $4.3 million, up from $3.1 million a year ago, reflecting continued investment in Gremlin-X and our AI software. General and administrative expense was $8.1 million, up from $4.2 million a year ago, driven by added head count from our November acquisitions, hiring to support new business opportunities and higher stock-based compensation expenses. Stock-based compensation expense was $2.3 million this quarter, up from $1.1 million a year ago, driven largely by executive equity awards approved by shareholders in June. This is a noncash expense and was a primary driver of the increased general and administrative expense and operating loss versus last year. Going into the third quarter, this expense will increase materially from the second quarter since we only recognized roughly one month worth of noncash expense in the second quarter related to the new awards. Sales and marketing expense was $2.3 million, up from $1.3 million a year ago, reflecting the expansion of business development efforts within our Palladyne Aerospace and Defense division as we pursue more opportunities with the Department of Defense. Operating loss for the quarter was $13.4 million compared to $8.1 million a year ago. Below the operating line, we recorded roughly $1.1 million of other income, which includes $800,000 from a noncash gain on the revaluation of our warrant liabilities, together with interest income of $300,000. GAAP net loss was $12.3 million or $0.27 per diluted share compared to $7.5 million and $0.20 per diluted share a year ago. On a non-GAAP basis, which excludes the impact of warrant revaluation, stock-based compensation and the change in our contingent consideration liability, net loss was $10.8 million or $0.23 per diluted share compared to non-GAAP net loss of $6.4 million or $0.17 per diluted share in the prior year period. Turning to liquidity, we ended the quarter with $43.7 million in cash, cash equivalents and marketable securities, flat with the first quarter and down $3.4 million from the 2025 year-end balance. Operating cash burn, which we define as cash used in operations plus CapEx, was $11.1 million during the quarter. This includes approximately $10.2 million of operating cash and $900,000 of CapEx. This is offset by approximately $10.8 million of net proceeds from our ATM program and $300,000 in interest income. During the first half of 2026, we raised $17.3 million in net proceeds under the ATM at an average price of $6.84 per share. We view the ATM as a flexible tool that we can deliberately use to expand our runway at a low cost of capital. As of June 30, we had approximately $24 million in remaining capacity under the program. Backlog ended the quarter at $24.6 million, an increase of 43%, driven by approximately $13 million of new contract awards during the quarter, partially offset by revenue recognized during the quarter. We expect a majority of our current backlog to convert to revenue over the next 12 to 18 months. Turning to guidance, we are reiterating our full year 2026 revenue guidance of $24 million to $27 million, which represents roughly 357% to 415% growth over 2025 revenue of $5.2 million, and we continue to expect organic growth across each part of the company. We're also reiterating our full year operating cash burn guidance of $32 million to $36 million, or roughly $8 million to $9 million per quarter on average. We ran above that pace in the first half, although we offset most of that with funds raised via our ATM, and we expect operating cash burn to trend lower through the second half as revenue and margins ramp. Based on our liquidity position, our access to the ATM and our expected backlog conversion, we believe we are well positioned to execute our plan. Operator, we're now ready to take questions.
Questions and answers
Our first question comes from Maxwell Michaelis with Lake Street Capital Markets.
Congrats on the quarter. A couple for me, a bunch of different topics here. First, I want to touch on the IAI. You guys mentioned it, but my real question around that is kind of what sort of progress has been made on the Americanization process. Have you identified any domestic manufacturing sites? Do you need that? Have you been in any preliminary conversations with the U.S. government just around these systems and kind of getting your foot in the door on that end?
Max, it's Ben. Thanks for the questions on IAI. We have been very active. We just announced the deal roughly seven to eight weeks ago, and we hit the ground running with kind of a full-court press, ticking down the list of every potential U.S. government customer that could be interested in the systems, and we have engaged now, I would say, with about half of those targets. Across the board, there is meaningful interest, but interest is just the beginning. Then you have to get into the specifics and details about performance and capability and all of those kinds of things. But the bottom line is we have yet to have anybody that we targeted tell us that there's no interest. So we are extremely encouraged and feeling bullish about the opportunity set. In many cases, there was existing familiarity with these weapon systems and access to them here in the United States. So we're not starting from a standstill position because these are battle-proven weapon systems that the community is generally familiar with. In terms of facilities and our capacity to manufacture, we have a fair bit of manufacturing capacity in the space today. The real question is what kind of volumes the government might want; we could quickly outgrow our capability in terms of size and facilities. So we are in the process of identifying additional facilities — not that we would need to build and construct because of the timelines associated with that, but we are developing a short list of currently available facilities that, if the government were to say, 'Let's go,' we'd be ready to jump in. And hopefully, without a lot of modifications, be ready to start standing up manufacturing in the relatively near term.
Awesome. Makes sense. A couple more from me. Secondly, Gremlin-X advanced from TRL 6 to TRL 7 in the quarter. What was that specific milestone that got you there? And what is required to get you to TRL 8?
Over the quarter, we did five different design iterations. We've talked in the past, Max, about how our team can move quickly. That was definitely moving quickly: five different design iterations, flying and testing in each instance. What got us over the hurdle was getting into the hands of customers and having the customers actually fly as intended — not us managing or demoing the systems, but actually having soldiers in the field using them under battlefield conditions. So we now feel like we're in a great spot with that. One could argue whether we're at TRL 8 yet or not, but because we haven't been in live-fire battlefield conditions we will hold off on making that announcement. But the bottom line is the systems performed as expected and ahead of schedule.
Perfect. Last one from me. I think you need $15 million to $18 million of revenue to sort of hit the guidance range for 2026, backlog is around $25 million that's going to be converted over the next 12 to 18 months. When we think about that last $15 million to $18 million in 2026, what percentage of that is already contracted to be shipped this year, if you could share?
So when we talk about backlog, that number is contractually committed, binding legal contracts. What is less certain is exactly the timing of deliveries and how they align with milestones and revenue recognition. So it's a little challenging for me, Max, to give you a straight answer on how much of that backlog will absolutely be delivered in 2026 before the end of the year. What I can tell you is we are feeling — based on everything that Trevor and I see in the business — confident in reiterating the guidance that we previously gave.
Our next question comes from Mike Latimore with Northland Capital Markets.
Great quarter. Great bookings. I guess on SwarmOS, can you talk a little bit about the exercises you're involved with and compare that to some of the programs that are visible like Swarm Forage? What's the difference between the exercise you're involved in versus, say, the Swarm Forage program? What are the different requirements? What are the relative opportunities? Maybe just help us clarify that a little bit.
Thanks, Mike. The military is a large, multifaceted organization with lots of different groups doing different things. The Defense Innovation Unit, or DIU, has a focus on finding and identifying companies with products and technologies that could be useful to the military. They frequently will stand up competitions where they put out specifications and requirements to industry, run a bake-off, down-select and then give some money to develop a product or technology into what they're looking for based on the best criteria. That's one path for a young or new entrant to get products and technologies in front of people at the Pentagon. There is another path: to have a finished product that's ready to go and to get it in the hands of actual users in real-world exercises. That is what we have done with both Ivy Mass and PC-C6. It's not about smart technologists and businesspeople doing a controlled test; it's actually getting product and technology in the hands of warfighters, letting them get real-world experience with it and evaluate it, and then developing a report that says to the operating units and the acquisition community, 'We've used it in the field; here's our assessment.' What we've announced this morning and in prior press releases is our success at real-world battlefield-condition exercises, not demos that we control. We train the teams that will use the systems, provide the drones, and they perform like they're in the field. We have no ability to control it once it's in their hands. These are generally soldiers who may or may not have drone experience and certainly not experience with autonomous systems, and part of what we get evaluated on is whether soldiers with limited training can use the system and accomplish the intended mission. We have been gratified that at every turn we've gotten great feedback. We're delivering what we said we would deliver, which results in getting invited to more exercises with diverse groups of potential users. Our reputation is spreading organically across different units.
Yes. Awesome. And then I guess as you look to the third quarter here, it seems like there could be a really heavy 'sweeps' sort of period here. I think you talked about spending more money through fiscal year-end. Does it feel like this quarter could be another healthy bookings quarter or even better than the second, given that?
It's premature to say definitively. When you talk about cleanup money that's available, it's a bit like watching a game of musical chairs: everyone in the defense industry is hoping to capture some of that. Much depends on who has access to what capital and each group's priorities. I'm hopeful, but I wouldn't say I have a high degree of confidence until we actually see the awards — until the music stops, you don't know who's got the chair.
Sure. Makes sense. Last one for me on IAI. They have those three products. Any sense of whether a customer would buy all three? Is there more interest in one versus the others? Any feelers there?
Different customers have different use cases and missions. For longer-range and extended-distance missions, customers may prefer the larger platforms like HAROP and HARPY. If launching from a vessel or for shorter-range missions, they might prefer the Mini HARPY. It's uncommon for a single customer to need all three because they generally operate in different environments with different purposes. That said, a large customer like the U.S. Army could potentially procure all three to cover a range of missions, but for specific operational units the mission sets typically bifurcate.
Our next question comes from Adam Samuelson with Jefferies.
I wanted to hone in on the cash burn and outlook for the balance of the year. You've got an outlook that has improved revenue on average in the second half each quarter. How do we think about that scaling into the cash line versus margin versus working capital and other operating expenses? I'm just trying to think about the kind of base cash spend underlying the business and how that can scale into cash flow prospectively on greater revenues as we think about 2027.
We reiterated our guidance for the year. If you do the math, to hit those numbers in the back half, cash burn is about $12 million that we're anticipating. There is some CapEx built into that that we'll use strategically. We do see, with revenue trending up and more margin dropping to the bottom line, that cash flow improves and cash burn requirement goes down. We have quite a bit of capacity, both from a manufacturing standpoint and from a product standpoint when we start getting into our AI software opportunities to be able to scale the business without having to add a lot of incremental costs. So we're optimistic about seeing that trend down and, with some planned investments in the business, being able to significantly reduce the cash burn in the second half.
Okay. And maybe just along those lines of the $24 million of backlog, can you share whether a proportion of that is actually software-related revenue or software licenses and so would presumably carry very high incremental gross margins?
The backlog composition is roughly, and this is a rough breakdown, about 50% product sales, which includes our precision components and our avionics. Avionics tend to be higher margin and precision components a bit lower. Roughly 30% relates to our software. I want to caution that most of the current software revenues are tied to trial contracts and development contracts with the military that have not yet transitioned into high-volume licensing. But it does show we have a meaningful amount of software-related backlog. It is a good-margin business, but it's not yet the very high gross margins we'd expect when we start selling high-volume software licenses. The balance of the backlog is in engineering services.
Our next question comes from Brian Kinstlinger with Alliance Global Partners.
With the successful results at Ivy Mass and PC-C6, can you talk about the next steps in the sales cycle as the end customer gets educated and sees these positive results? How should we think about these events turning into production orders?
There are two parts. For software, reports from these exercises will be aggregated and reviewed by those writing requirements and by purchasers. As demand for autonomous and swarming solutions increases, solicitations will likely be released that more closely resemble what is being trialed in the field with our software. Then we respond to those solicitations and hope to be down selected for contracts. We've also been asked about retrofitting our software onto drones already deployed to units in the field; the numbers discussed are sizable and present logistical questions about distribution and installation. That shows the level of thought going into delivering this capability broadly. On the Gremlin-X hardware side, I believe in the relatively near term we'll start getting inquiries about volumes we could deliver in terms of time and cost. My expectation is those volumes will start modestly — hundreds of units rather than thousands — but awareness of the need for low cost-per-effect mini bomber solutions is growing. So I think we're at the cusp of seeing volume opportunities in both software and hardware.
You highlighted you have 20 proposals submitted with several hundred million dollars in potential. Is that mostly hardware? It sounds like the software procurement may be a bit further out. How should we characterize that proposal activity?
Most of the larger dollar opportunities are combined hardware and software solutions. One of the advantages we bring is the ability to jointly develop hardware and software that are optimized together — sensor, compute and AI — which delivers a different category of capability to warfighters. So a majority of those proposals reflect combined complete systems incorporating both our hardware and software.
Our next question comes from James Kisner with Water Tower Research.
Congratulations on PC-C6. That event moved rapidly from the rehearsal to a larger event. What did you learn from operating in that faster, bigger environment in general or about SwarmOS?
Some key takeaways: we confirmed that a single soldier can focus on mission-level tasks and not have to micromanage each drone. While we knew it was possible from a technology perspective, seeing soldiers with limited drone experience use the system validated that it works in practice. We saw no limitations on the number of drones a fairly inexperienced soldier could manage, which is encouraging for scaling. We also validated performance in degraded or challenged communications environments — a critical real-world condition — and the system performed as designed. We learned how to best train new users, compressed training time so more people could use the system faster, received specific UI feedback and met a short timeline challenge to integrate with another company's command and control system, which we accomplished. These field lessons reinforced our assumptions and provided direct feedback from users.
That 30-minute training time is pretty impressive. More broadly, it seems the hardware-independent multi-vendor approach resonated at PC-C6. As the Army moves towards larger fleets of autonomous systems, how durable an advantage is vendor neutrality versus a closed single-platform ecosystem?
The Pentagon's current mantra emphasizes open architecture and avoiding vendor lock. That aligns well with our approach. Real-world operations often involve units using drones from multiple vendors; enabling cross-vendor collaboration addresses an important capability gap. We believe the military will prefer not to be beholden to a single provider. Our vendor-neutral software approach is a durable advantage because it allows disparate platforms to interoperate and collaborate, offering significant operational value.
You've integrated a number of manufacturers' aircraft, including Anduril. Is bringing a new platform on board getting faster and more repeatable?
We've learned a lot and the process is getting more repeatable, but integration complexity still depends on the OEM and the drone architecture. We've integrated mostly rotor platforms to date; we're now integrating our first fixed-wing platform, which has different altitude, sensor and compute characteristics. Integration timelines vary; based on current knowledge, it can range from two to four weeks with a couple of engineers working full time to complete an integration. We're better at asking the right questions up front and understanding what needs to be nailed down before beginning the process.
Our next question comes from Ralph Weil.
Good results. Are you involved or trying to become involved at all in the Golden Dome program? If so, which of your products would fit into that? And second, your STRATFI program is now at $10.6 million. What is the potential market for that product used there and where will your capabilities be used?
On Golden Dome, the primary focus today is on larger, more exquisite weapon systems to intercept large incoming threats. There could be opportunities for us at more localized scales to deal with smaller threat form factors like swarms of smaller drones, but Golden Dome's priority right now is larger systems, often leveraging space-based assets, which is not our domain. As that effort evolves and attention turns to regional domes addressing smaller weapons, we may have an opportunity to participate. Regarding the STRATFI contract, that relates primarily to our IQ product, which is our industrial autonomy platform. IQ focuses on tasks that are difficult, dangerous, repetitive and hard to staff, such as surface preparation and paint stripping for large aircraft components. That's the work we're doing with the Air Force and the $10.6 million program. The total market for substituting automated and autonomous machines for these kinds of tasks — both military and commercial — is very large, in the billions of dollars globally. Our approach is to go deep in individual verticals where we've seen success and then expand: land in a focused use case and expand from there. We introduced version two of the software earlier this year, have strong engagement with customers and systems integrators, and expect more placements in the second half of this year.
That does conclude our question-and-answer session. Thank you all for your participation today. This concludes today's teleconference. You may disconnect your lines, and have a wonderful day.