Good day, and welcome to Zapata Computing Holdings Incorporated Second Quarter 2024 Financial Results and Business Update Conference Call. As a reminder, this conference is being recorded. It is now my pleasure to introduce your host, Edwardo Rojas. Thank you. Mr. Rojas, you may begin.

Thank you, Kat. On today's call are Christopher Savoie, Chief Executive Officer and Co-Founder; Sumit Kapur, Chief Financial Officer; and Jon Zorio, Chief Revenue Officer of Zapata AI. Earlier today, Zapata AI issued a press release announcing its second quarter 2024 results. Following prepared remarks, we will open up the call for questions. Before we begin, I'd like to remind you that this call may contain forward-looking statements. While these forward-looking statements reflect Zapata AI's best current judgment, they are subject to risks and uncertainties that could cause actual results to differ materially from those implied by these forward-looking statements. These risk factors are discussed in Zapata AI's filings with the SEC and in the release issued this morning, which are available in the investor section of the company's website. Zapata AI undertakes no obligation to revise or update any forward-looking statements to reflect future events or circumstances, except as required by law. With that, I would now like to turn the call over to Christopher Savoie, CEO and Co-Founder of Zapata AI. Christopher?

Thank you, Eduardo, and good day to all. We saw strong momentum across various fronts during the second quarter of 2024. As you'll hear more about during this call, this is evidenced by the step-up in revenues and gross margins versus Q1 2024 and the year-ago quarter, the expansion of existing and the formation of new strategic partnerships, and the healthy growth we saw in our qualified pipeline of prospective customers, which today stands at more than $30 million. Before elaborating on these highlights, we think it is critical to touch on the bigger picture and, most importantly, how Zapata AI is carving out a compelling niche in the industrial AI marketplace. Understanding of generative AI, its applicability, its practicality, and the business model and economics behind it is evolving daily. Enterprise users of this technology are coming to appreciate that there can be different approaches for different applications. Let me elaborate. As disclosed in a report published by a prominent investment banker earlier this summer, generative AI can be very expensive. As noted in the report, the tech sector is poised to spend more than $1 trillion on generative AI CapEx in the coming years. Similarly, a leading venture capital firm evaluated AI investments and calculated that the entire industry needs to generate $600 billion annually to break even on its initial expenditures. The root of these high-dollar figures is technology centered around large language models or LLMs. There is, in our mind, no doubt a large market for LLMs, but LLMs are not a be-all, end-all when it comes to generative AI and industrial AI. At Zapata AI, we firmly believe that the value of generative AI is not going to come exclusively from one large model that rules the world. Rather, there is a vast market for an approach to AI for enterprises and customers that is based on ensembles of highly tailored, highly specific, and relatively small models. This is the approach we have been taking at Zapata AI from day one. LLMs are limited in their ability to solve truly complex business problems, especially those involving the handling of structured numeric or time series data. But large enterprises in the Fortune 500 have mission-critical use cases that involve numbers and time series data, not just text. Small, precise models can tackle these deficiencies, and Zapata helps businesses build and optimize these specialized models to reduce costs and increase revenue. Further, our small ensemble model approach allows us to address high CapEx hurdles for training and deploying AI. Our quantum-informed mathematical models are incredibly efficient and do not require vast amounts of compute, unlike traditional LLMs. As we've all now seen, compute can be extremely expensive and is largely inaccessible given the limited chip supply. Further, the energy models needed to power these compute-intensive LLMs are significant. Lastly, as you've heard from me before, but it bears repeating, we build applications using our customers' data on our Orquestra platform, our software platform, which can be deployed on any customer's cloud in their own secure environment, including on the edge. This is key to avoiding vendor lock, which we've been hearing from customers is a major concern, while removing data governance, security, and privacy concerns. Our leadership team, global business development staff, and world-class engineers are out there speaking with prospective customers every day, delivering this message, and we routinely see the proverbial light bulbs going off. Over the past several months, and especially in the few months since we closed our acquisition with Andretti Acquisition Corp., we have seen increased interest in how we can deliver value to customers with our small model ensemble approach. The business development and sales discussions we are having, in many instances, with very large organizations in the Fortune 250, are the mainstay of what we're doing. Relationship building and decision making are time-intensive processes that involve alignment across multiple facets of an organization. We remain incredibly optimistic that these conversations will continue through the typical sales conversion cycle and that we will convert several of these exciting opportunities into commercial agreements as we get deeper into the second half of this year and beyond. Once we are in with a customer, the relationships we have established are incredibly sticky, and the opportunities to diversify and scale within an organization are tremendous. This is well evidenced by how we've expanded our partnership with Andretti Global and AI by expanding use cases beyond race strategy into digital transformation and operational optimization. The same goes for our DARPA relationship, which goes back to 2022 and has expanded over time. As we look forward, we are optimistic about further scaling up and expanding revenue per account with our existing customer base. Against this backdrop, I would like now to introduce Sumit Kapur, our CFO. At the time of our last call in mid-May, as you may know, Sumit was just coming on board. After a few months in the seat, he is well-positioned to talk in more detail about the process and platform that we have, in essence, how we bring our customers up to speed on our compelling value proposition. Sumit?

Great. Good morning, everybody. Thanks, Christopher. Great to be here today and to speak with you all for the first time. As Christopher suggested, we believe the AI market is moving to us. At Zapata AI, we are able to guide customers as they progress along their AI journey. Let me emphasize, we very much believe this is a journey, and the right AI tool is not off the shelf. Those of you familiar with the world of software have likely heard the term Software Development Lifecycle, or SDLC. SDLC is a methodology used to define, design, test, integrate, and deploy software to ensure the best end product possible for the customer. At Zapata, we run Model Development Lifecycle, or MDLC, tools on our Orquestra platform. This MDLC platform, combined with our quantum AI-inspired model library, ensures that we guide customers through their AI journey efficiently and sustainably, resulting in better business decisions powered by some of the world's most advanced AI models. This MDLC process starts with discovery, a process of understanding the customer's operational challenges, their priorities, and what we are solving for together. We then move on to data engineering, model selection, and prototyping—that is, determining the models and applications we need to build for the problem we are addressing. From here, we move on to critical model training and benchmarking—that is, building the best fit bespoke models and applications based on the customer's data. And lastly, we build and deploy the application and then monitor, maintain, and improve it, effectively working side-by-side with our customers to ensure that they have the ability to make more intelligent and optimal business decisions based on the AI technologies that we have jointly developed. We are today doing this through two complementary business lines. Our scientific solutions business line works with innovation centers at government agencies, academia, and corporations to conduct research that leads to new technology, models, and approaches. In this way, this business line generates not only revenue but also intellectual property. On the other side, our commercial solutions group leverages this IP to develop and deploy AI applications that provide practical business solutions for our commercial customers. Over time, the more models we build and deploy, the more development we do on our platform, and the more customers we work with, the faster we expect to bring in new business. We're already seeing customers start at different points of the maturity scale, but as everyone learns how best to leverage AI, the learning curve will flatten and new customers will come in having a better understanding of how to apply our MDLC platform and hyper-efficient AI models. Finally, I'll note that as this progression along the AI journey occurs, our gross margin percentage with each individual customer increases. All of this supports a growth outlook for our business that is anything but linear, and we are truly excited about the various opportunities we see coming down the pike. To talk more about this and our Q2 highlights, I'll now turn the call over to Jon Zorio, our Chief Revenue Officer. Jon?

Thanks, Sumit, and good morning to everyone. I'm excited to talk today about the progress we've made in building our sales pipeline since the beginning of the year and to provide some updates on our other go-to-market and commercial developments from the second quarter. As of today, we've identified more than $30 million in qualified sales opportunities based on the ongoing and advanced-stage customer conversations we're having. As Christopher mentioned, we're happy to report that the customer interest in Zapata and our differentiated AI solutions continues to grow, and this is reflected in our rapidly expanding pipeline. We've taken a balanced approach in driving both direct sales and partner-led pipelines. We've launched our executive ambassador program, and we're gaining early traction. We've also begun to see top-of-funnel and marketing qualified lead activity pick up in recent months as our awareness and thought leadership campaigns have scaled up and are making an impact. These advanced-stage opportunities cut across our core verticals, which, as we discussed in last quarter's call, include financial services and insurance, telecom, media, and technology, industrial and advanced manufacturing, biotech and pharma, and defense. The opportunity pipeline is relatively well-balanced, with no one vertical clearly dominating the others, which speaks to the broad-based resonance of our message. I'll highlight several examples of significant progress made during Q2 on these various dimensions. Our engagement with KPMG has begun to show positive early results, specifically in the financial services and insurance verticals. Through an accelerated proof of concept from one of KPMG's long-standing global life insurance clients, we were able to demonstrate our ability to transform risk and compliance modeling at scale. Through our advanced optimization techniques and world-class data engineering expertise, Zapata was able to reduce the time and effort required to run a massively complex compliance model by more than 1,000 times. That's literally a reduction of thousands of man-days while reducing the level of effort by 90% and maintaining model accuracy. In defense, we've achieved yet another significant milestone as a lead performer in our third year of work with the U.S. Defense Advanced Research Projects Agency, commonly known as DARPA. Along with our program partners, Zapata published our research findings at a critical milestone in DARPA's quantum benchmarking program to quantitatively measure the value of specific transformational quantum use cases and the associated hardware required. As the only company participating across all program tracks, Zapata will execute on the remainder of Phase 2 for 2024 and well into mid-2025. Also in the defense sector, yesterday we announced a Cooperative Research and Development Agreement, or CRADA, with the U.S. Special Operations Command. SOCOM is a unified combatant command responsible for all special operations units within the U.S. military, including elite forces such as the Navy SEALs, the Army Special Forces, otherwise known as the Green Berets, etc. In support of its hyper-enabled operator and hyper-enabled force initiative, Zapata will be working with SOCOM teams to enhance situational awareness, real-time decision-making, and operational readiness in challenging environments and contested spaces. We'll be accelerating SOCOM's ability to plan, develop, and deploy AI-driven advantages to align closely with specific mission objectives and parameters. The decision support capabilities developed by this partnership will run on the edge in low or no connectivity environments and will be optimized for ruggedized high-performance computing hardware. We continue our pioneering work with D-Wave, a leader in quantum computing systems, which has led to the discovery of novel potential cancer drug candidates. During the second quarter, we are pleased to expand our partnership with D-Wave. Our expanded collaboration will work to speed the development and delivery of advanced AI systems, which combine quantum, quantum hybrid, and classical GPU computing. In our last call, we discussed our partnership with Tech Mahindra, a leading global systems integrator. We continue to advance client discussions with Tech M's telecom client as we establish use cases to transform network operations through industrial AI based on anomaly detection to predict service disruptions before they occur, including on the edge. This, in turn, can deliver greater uptime, lower OpEx, superior customer satisfaction, and broad operational efficiencies. In another exciting development in Q2, Zapata joined the KT Consortium. This consortium brings together leading industrial companies in chemical and biochemical manufacturing, such as BP, Henkel, Mitsubishi Chemical, Total Energy, and Arkema, with advanced technologies in academia. Zapata AI aims to work closely with these consortium members to streamline the development and manufacture of complex chemical compounds by leveraging our Orquestra platform proprietary code library and quantum-based algorithms. Our participation in the KT Consortium builds on our history of successful collaboration with BP and BASF in the chemical space. To conclude our update, Zapata is also exploring opportunities to monetize our expansive IP and patent portfolio. We see a highly differentiated and potentially lucrative opportunity to help enable the next generation of AI, which will rely on end-to-end quantum software and hardware systems. Many hyperscalers and the largest names in tech are already beginning to consider the applicability of quantum systems as the next evolution of AI, given its insatiable resource requirements. We'll be sharing more about this strategy in future calls. I'll now turn the call back to Sumit to review our second quarter financials.

Thanks, Jon. I'll dive into the financials. Q2 2024 revenue was $2 million, which compares to revenue of $1.43 million in the year-ago quarter and $1.22 million in Q1 2024. The period-over-period increase reflects increased software license delivery. Gross margin in Q2 2024 was 36%, significantly better than 19.7% in the year-ago quarter and the 13.6% we reported in Q1 2024, as we advance our ability to drive greater profitability with increases in scale and individual customer progression along the AI journey. Operating costs during the period were $8.09 million versus $4.42 million in the year-ago quarter and $5.24 million in Q1 2024. General and administrative costs drove the majority of the increase relative to both periods. A significant portion of this change is due to one-time fees that resulted from the company's NASDAQ public markets listing and the filing of a registration statement on Form S1 in connection with the company's equity line of credit, which drove higher professional services fees. Our GAAP operating loss was $7.37 million in Q2 2024 versus $4.14 million in the year-ago quarter and $5.08 million in Q1 2024. Our GAAP net loss during Q2 2024 was $15.58 million. The variance to our operating loss primarily reflects $8.23 million of non-cash expenses related to the fair value change of a forward purchase agreement derivative liability. Turning to our balance sheet and cash flow statement, on June 30, 2024, we had $7.16 million in cash and cash equivalents, including $0.14 million in restricted cash. Net cash used by operating activities was $6.11 million during the second quarter of 2024. Included in this figure is $0.89 million in cash generated by working capital. During Q2 2024, we raised a net total of $6.06 million through financing activities, of which $5.30 million was raised through our equity line of credit with Lincoln Park Capital. Subsequent to the end of Q2, we also entered into an additional purchase agreement with Lincoln Park for up to $10 million in shares of common stock subject to certain conditions. As we stated last time, we will be practical, judicious, flexible, and opportunistic in our funding strategy. That concludes our discussion on financial results. Back to Christopher for closing thoughts.

Thanks, Sumit. These are incredibly exciting times for Zapata AI. As a business and as other businesses look into industrial AI to solve for specific mission-critical business use cases, our unique small and ensemble model approach is able to drive a return on investment in relatively quick fashion. We're incredibly excited about our $30 million-plus pipeline and we look forward to making headway on converting these opportunities, while further growing this figure across our four verticals. At the same time, we will continue to advance and expand the partnerships we already have in place while also exploring potential ways to monetize our IP and patent portfolio, as Jon alluded to. Thank you all for your time and attention, and we look forward to updating you on our next progress on the next quarter call. Thank you. Operator, over to you for Q&A.

Thank you. We will now be conducting a question-and-answer session. Our first question comes from Yi Fu Lee with Cantor Fitzgerald. Please proceed.

Thank you for taking my question and congrats, Jon on the strong momentum in the second quarter. Maybe one question for each, for Chris, Sumit, and then Jon. Chris, in terms of definitely a lot of projects going on in the pipeline, you've got KPMG, Tech Mahindra, DropUp, Phase 2, and then yesterday you won the special ops award. I was wondering if you could rank order for us in terms of these opportunities you see in the pipeline. What is the most material that we should look into to stay focused on and from the lessons learned from these pipeline generation, what do you think are the next areas you're going to focus on?

Yes, thank you for that question. As Jon alluded to in his comments, it's pretty well balanced. So it's not like, okay, you have to look at this sector more than that. I'm really excited about the defense work. Long term, as you've seen with other companies like Palantir and C3 AI, that is a very sustainable, sticky, and great revenue source. It's really important that I think strategically that we double down our efforts in that field. We have some really important technologies that we've tested on the racetrack on the edge with the AI that has equal applicability for unmanned vehicles, drones, satellites, and other moving vehicles that the defense department cares a lot about. They're equally applicable, though, in the commercial space, and it's very interesting. I think we commented on the last quarterly call about how the time series data that we're doing in racing is equally applicable to time series data in trading with the work that we're progressing with Sumitomo Mitsui Trust Bank and others. That kind of work is applicable. You've seen the work we've done in insurance as well in the financial sector. So, financial sector, I think that was pointed out as a key growth area. Some of the stuff that we're doing in time series is extremely applicable there. The efficiencies that we're getting, a thousand times better efficiencies, and using GPUs is really amazing and really applicable, driving actual ROI for finance. And then across other industrials, we're seeing the same kind of thing with the time series AI applications in particular. So it's pretty well balanced. I'm really excited about finance; I'm really excited about defense, but with the other companies in the KT consortium, there are many other industries that do this. I think it just speaks to the real generalizable applicability of this in industrial situations.

Okay. Thank you for that, Chris. And then moving on to Sumit, extremely good momentum on the revenue side and then obviously gross margin not at the 60% scale. Maybe, Sumit, can you help us frame like it's a 40% growth on the revenue and you already got 16% upside on the GM? Should we get used to this, I guess, incremental uplift? And anything to help us guide 2024 would be helpful.

Thank you. Great to see you here. As before, we are not yet at a point where we're providing specific forward guidance, but what I will say is just to underscore the point that was made earlier, which is there are some key dynamics that are driving this. Our penetration into the market—the more we penetrate into the market and the more we demonstrate these use cases, the more interest we see and the more ability we have to close pipelines. So that's definitely positively impacting our ability to close pipeline and increase revenue. And the other point that was made a little bit during our comments, but I really want to emphasize again, is that the model development life cycle, or MDLC approach that we take, has multiple stages. And you can imagine that at earlier stages, such as discovery and data engineering, growth margins are lower than at more advanced stages, such as deployment, monitoring, and management. So the growth margin actually increases. And at the same time, synergistically, the ROI for the customer increases as you advance along the customer AI journey. So built into the model are advances in our ability to accelerate revenue and accelerate growth margins. But other than that, at this time, we do not have any other specific forward guidance.

Got it. So the more mature projects, a better margin. Just like before I move on to Jon, is there any other charges like one-time charges that we should be aware of in the second half? I know you have the SEC, the standard SEC filing regulatory fee this quarter. Anything that we should be aware of in the second half?

In the second quarter, really the big charges were associated with going public. So it was professional fees associated with that first queue. And so not only the ELOC-related charges but also going public. And so there's a number of professional fee charges that are one-time.

How about like the second half, meaning like the third quarter? Any other, I guess, standard charges?

Yes. Sorry. In the second half, no, things should moderate considerably.

Okay. And then lastly, flip over this to Jon on the go-to-market side, $30 million from Pipeline, congrats on that. In terms of the timing to monetize this $30 million worth of Pipeline, understand on the footnotes, it says that it's budgeted, but you still have to go for the bid. Does it mean, Jon, that you still have to, I guess, like bid with other firms, other competitors in this space to win that $30 million? And when do you think it's going to translate into hard revenue and cash flow? And lastly, Jon, in terms of your hiring needs for the second half, can you also comment on that? And that's it from me. Thanks very much.

Thanks for the questions. I appreciate you being here. I'm going to try to take those in the order you asked them. So our typical Pipeline dynamic is these are enterprise deals. So the Pipeline, or rather, the deal itself cycle could play itself out over a typical 12-month cycle. So those are, within that $30 million Pipeline, those are at various stages of maturity, as Sumit alluded to, based on the MDLC stage, if you will, where those deals are. Now, those are qualified deals that we've established a budget, where we've established a need, we've established a timeline on the client side. And so we push hard to move these deals along in terms of conversion as we move through our proof of concept stages and our deployment stages. And so we work very closely with our customers. And part of one of the earlier questions was, what's been resonating? I think what's been resonating is our approach, frankly, is we've been very hands-on and consultative moving clients through the MDLC as we're going through this new process for some of these customers every step of the way. And so we work hard, but we have a very structured approach, and it's very clear to the client in terms of what comes next, what comes after that, and we're kind of taking it through that approach in a very organized manner. So there's no surprises, and they're seeing the value at each step along the way. But those will play themselves out over, let's say, within the next 12 months. From a hiring standpoint, we've been very pragmatic in terms of adding sales and BD and marketing resources along the way. So we've been hiring kind of as demand has dictated. We've been hiring in places around the world where the customers have taken us. And so we're going to maintain that sort of disciplined approach.

Okay. Thank you for that. Thank you, guys. I'll be back in the queue.

Our next question comes from Matthew Harrigan from Benchmark. Please proceed.

Just one LLM to rule them all out of the ranks. I have one question and then one follow-up. You've got a vast opportunity, to say the least, and you're able to demonstrate the superior quantitative efficacy of your solutions for computation, energy consumption, time to market, and all that. Clearly, even though you've got great IT, you're constrained in terms of your ability to immediately meet the opportunity that looks pretty urgent across many verticals. And clearly, you're looking at licensing IT and all that. You've got a number of ways to address that. But what are the risks that you get more competition along the way? Are you seeing anything happen latent or otherwise that concerns you where you might not be able to develop the huge opportunities that you have?

Yes. Thank you for that, Matthew. This is Christopher. I think, yes, obviously, there's going to be competition. I think that that's a good thing. If there were no competition, there probably isn't opportunity there. I think what we're finding is that it's really the legacy players who are in an enterprise already, the incumbents in data analytics that we find in the same lobbies as the customers, the usual suspects, the Microsofts of the world, the Palantirs, the C3.ais. I think we compete favorably there. We've competed with incumbents in the past and won accounts. And as you can see with DARPA, we were re-upped in some of those big names. We're now in the second phase of that project. And we're the only vendor in DARPA to be in all of the programs across that program. So I think we punch way above our weight in these accounts. But yes, it is the incumbent analytics companies that are showing up that are our competition. I think there we do have a newer tech stack that emerged from Harvard in 2017. We have the quantum mathematics capability advantage that I think we're recognized as having, par none, in the field and recognized as such by DARPA and others. I think we also, we've been nimble and able to actually get to deployment on these systems. And I think that's significant. It's one thing—and I think you're seeing this out in the marketplace—it's one thing to talk about AI, to kick the tires, to do a proof of concept, but it's another to get these things into actual production with real data and provide real ROI. And that's where the focus is going now on the market. So I think we stand in a very favorable place vis-à-vis that competition, say older, bigger competition in closing these accounts.

Yes. As a scientist, you sound surprised by the applicability of verticals you're seeing homogeneity modeling the path of diversity in energy companies. What do you think is challenging for that?

Yes. The downside of the diversity of customers is that there's a diversity of customers. For us, though, the good news is that the math is all the same. Like I said, the work that we're doing, the actual model technology and the process, the MDLC process, the model development life cycle that we bring in our tools with Orquestra, all the platform is the same. A lot of the technology, a lot of the math is exactly the same, whether you're on a racetrack or you're in a banking trading application, an insurance application, or a defense application. So that's the kind of beauty of the platform is that it's broadly applicable. I mean, the downside and risk is you can't do everything all the time. It'll take time to grow. It takes time to close any of these accounts. These are multi-million-dollar, multi-year deals. And so we've been very specific about certain verticals and concentrating there so that we can get some efficiencies. This will help us, as alluded to in our gross margin, because as the tools and the models and that kind of capability builds up in Zapata, we're able to deliver that at a higher gross margin, but at a higher efficiency rate and quicker to our customer base.

And then complementing the deep science character of DARPA, you have the coolest press release ever probably with the U.S. Special Operations Command yesterday. Would you be likely to work with some of the other NATO countries or Iran or Israel on that? I mean, that seems like something that would be pretty readily accessible. I don't know how large a revenue bucket that would be, but I would think you'd have a lot of people's families when you look at the CNN or Fox News?

Yes, absolutely. These are particularly sensitive conversations, but obviously NATO alliance countries are in need of these systems. And there's a lot of interest because of our exposure to it, our credibility with DARPA and now the CRADA. Yes, absolutely. NATO countries are our place, and we do have sales folks and presence in Europe now. So yes, that is a part of our strategy and growth strategy is to fully deploy these assets in the greater NATO framework. Thank you, Matthew.

This concludes our question-and-answer session. I would like to turn the floor back over to Christopher for any closing remarks.

Thank you very much. I really appreciate it. Thank you all for your time and attention. We look forward very much to updating you on our next quarterly call. Thank you very much for following us. Back to you, operator.

Thank you. This concludes today's teleconference. We thank you for your participation. You may disconnect your lines at this time.