EVgo, Inc. (EVGO) Earnings Call Transcript & Summary

Good afternoon. My name is Krista, and I'll be your conference operator today. At this time, I would like to welcome everyone to the EVgo Financial Modeling webinar. [Operator Instructions] I would now like to turn the conference over to Heather Davis. Heather, you may begin your conference.

Good afternoon, and welcome to EVgo's Financial Modeling webinar. My name is Heather Davis, and I am the Vice President of Investor Relations at EVgo. Joining me on today's call are Badar Khan, EVgo's Chief Executive Officer; and Olga Shevorenkova, EVgo's Chief Financial Officer. Today, we will be discussing EVgo's financial model followed by a Q&A session. Today's call is being webcast and can be accessed on the Investors section of our website at [email protected]. The call will be archived and available there will all with the presentation discussed during this call after the conclusion of the call. During the call, management will be making forward-looking statements that are subject to risks and uncertainties, including expectations about future performance. Factors that could cause actual results to differ materially from our expectations are detailed in our SEC filings, including in the Risk Factors section of our most recent annual report on Form 10-K. The company's SEC filings are available on the Investors section of our website. These forward-looking statements apply as of today, and we undertake no obligation to update these statements after the call. Also, please note that we will be referring to certain non-GAAP financial measures on this call. Information about these non-GAAP measures, including a reconciliation to the corresponding GAAP measures can be found in the presentation for this call, which is available on the Investors section of our website. And now I'll turn the call over to Badar.

Good afternoon. Thank you for taking the time to learn more about EVgo's financial model. As you know, EVgo is one of the leading electric vehicle charging companies in the United States, focused primarily on owning and operating DC fast charging. Today, EVgo has over 3,000 stalls in operation, including eXtend stalls at over 950 locations across more than 60 metropolitan statistical areas and in over 35 states. An investment in EVgo is not just an investment in sustainability, but also the investment in a market that has a multi-decade growth trajectory and a business model that delivers compelling unit economics. In our call today, we will take you through our proprietary network planning and underwriting process that we believe has already resulted in a competitively advantaged network of 3,000 operational stalls in better locations than our peers. The network planning process, together with our site development and construction capabilities, customer capture and retention track record and key partners and OEM relationships are together, the key components of the growth engine that is hard to replicate at the scale and with the financial returns we are now delivering. We had a great fourth quarter in 2023 and for the full year, we delivered record levels of throughput and revenue near trebling year-over-year. And with the operating leverage in the business, we are well on the way to adjusted EBITDA breakeven next year. Our network planning model has a rigorous but also conservative underwriting process. targeting double-digit returns on invested capital. This financial stewardship is key to the significant shareholder value that we expect to deliver over the coming years. With that, I'll hand over the call to Olga, who will take you through all of this in some detail.

Thank you, Badar. EVgo is an industry leader in its approach to network planning and underwriting. We have developed our own proprietary network planning and underwriting process, which we believe is unparalleled among our peers and is rooted in over a decade of data and insights we have accumulated about charging stations, their locations and their performance. The goal of our network planning framework is to maximize the cash flow we realized from every stall in the ground. Our network planning and underwriting is underpinned by sophisticated proprietary forecasting and special AI-driven algorithms that take into account grant availability, regional EV adoption and throughput demand, utility cost regimes and our host partners location. We have a dedicated team focusing on continual improvement to this process and the underlying algorithms and model. Let me walk you through our network planning and underwriting process, step-by step to illustrate its level of sophistication. Step number one, total demand for DCFC charging define the serviceable addressable market for EVgo. In order to forecast this demand, we look at projections of electric vehicles and operations or VIO, by metropolitan statistical area or MSA, and assumptions related to driving and charging behavior such as vehicle miles traveled, vehicle efficiency and percentage of charging needs delivered by DCFC in addition to multiunit dwelled density and rideshar presence in the geography. Step number two. We then developed multiyear forecast of the total number of seat stalled forecasted to be built in each MCA informed by third-party forecasts and our own views of which MSA's competitive dynamics and financial viability. Step 3. Then we develop utilization curves for stalls in each MSA based on our assessment of supply and demand. The curve for each MSA results in the forecasted mature utilization rate. That number is in the low to mid-20s for most MSAs we target. We conservatively do not assume that utilization will grow beyond that, even though we expect favorable market dynamics to continue. Said otherwise, EVgo does not assume that utilization levels need to be higher than the low to mid-20s to make the business case work. It is worth clarifying that utilization and throughput per stall, a function of the total demand for fast charging, divided by the total supply of fast chargers at the local level. And step number four, then we determine how much market share EVgo seeks to achieve in each of the MSAs that we target. Those assumptions are driven by EVgo's current position in which MSA, the MSA's financial attractiveness going forward and market-specific competitive dynamics. This informs EVgo's planned stall allocation by MSA. And then based on this analysis, the network planning team creates so-called bubble. The set geographical areas within our target MSA with site developers to focus on to find sites to develop. These bubbles are 2 miles in radius and have specific criteria that are calibrated to the geography that might be achieved in order to align to the business case. For example, target CapEx per stall or 30C eligibility. These bubbles are placed to optimally serve the demand for charging in that particular MSA, which is calculated at the sense of blog group level. The bubbles are designed to follow demand without cannibalizing any other EVgo chargers. These bubbles are also staged out over multiple years to ensure we do not get ahead of demand in any MSA. This process allows us to scale without sacrificing our rigorous underwriting standards. Site developers received clear instructions for each MSA and potential sites in each MSA are access continually through the process to ensure they are built according to our network plan and are set to meet target investment hurdles. This analytical framework is being constantly refined and updated with new data becoming available. The chart on Slide 3 shows our actual forecast accumulative throughput versus what we had originally forecasted for all owned sites and the outperformance demonstrates that our network plan and underwriting process produce differentiated outcomes, maximizing cash flow per stall. We underwrite our network to 10-year low double-digit pretax unlevered IRRs. Our underwriting model has been transferred out of Excel into Phyton, allowing us to run all kinds of sensitivities to our portfolio or certain parts of our portfolio within minutes. Our forecasting approach is fairly sophisticated, utilizing therefore, mentioned utilization curves, charge rate projections here and specific technical characteristics of the equipment being placed in the particular side location-based pricing, fairly precise modeling of energy costs using 24-hour load curves, peak coincident assessments, transformation and standby losses. Our model is also quite granular on the cost side, including all items in charge of network cost of goods sold, sustaining G&A and gross G&A, which is treated as one-off upfront investment alongside CapEx. Corporate G&A is the only cost category that EVgo has that is not included in the underwriting financial model. Let me walk you through each of these cost categories that were shared on our Q4 earnings call a few weeks ago. To increase visibility into our business, we have broken out our adjusted cost of sales into 2 categories: charging network cost of sales and other cost of sales. Charging network cost of sales includes all costs associated with running our core owned and operated charging network, whereas other cost of sales includes cost of sales associated with our expand and ancillary business. As of Q4 2023, within charging network cost of sales, approximately 40% of costs are stall dependent, in other words, fixed per stall. Stall-dependent costs include the majority of maintenance expenses, rent, property taxes, energy demand charges and connectivity costs. The other 60% of charging network cost of sales is throughput dependent and include volumetric energy costs, credit card processing fees and the variable portion of maintenance expense, which is a minor portion of total maintenance. Adjusted general and administrative expenses at EVgo can be broken down into 3 categories. First, the cost to sustain the existing charging network and other existing businesses such as coal center, third-party IT, account management, marketing, sustaining software and hardware expenses. These costs represent roughly 30% of total adjusted G&A. They are expected to grow as the network grows, although there is plenty of room to realize efficiencies over time with an increased network size, which we will discuss later. Second, the cost to support growth of the charging network and other business lines. Our so-called growth engine, we have spoken at length about during our most recent earnings call. These costs represent another roughly 30% of adjusted G&A. And finally, corporate overhead costs that represent roughly 40% of total adjusted G&A. Gross costs are tied to how stalls we're adding a year and corporate costs are fairly fixed at this point. Now let me do a deep dive into the economics of a single hypothetical stall that underpins our business case and clearly illustrates the attractiveness of EVgo's business model. Let us start with revenue. Annual revenue for any stall on our network is calculated as daily throughput per stall multiplied by 365 days, multiplied by average revenue per kilowatt hour. Average revenue per kilowatt hour in Q4 2023 was $0.57. This is calculated using reported charging network revenue of $28.3 million over reported throughput of 50 gigawatt hours. This includes revenue from both our retail and fleet customers as well as revenue from the sale of regulatory credits which are tied to throughput in California. We do not expect average revenue per kilowatt hour to change in real dollars over time. In other words, we assume our pricing will follow inflation from now on. Staying at throughput per stall is a function of both charge rate and utilization. Charge rate is the speed at which the individual vehicle can take electricity from the charter. Charge rate depends on both the vehicle battery, new cars with better batteries who have higher charge rates and the capacity of the charger. As we have shared previously, we are now mostly deploying high-power 350-kilowatt chargers across our network. For instance, some of the early electric vehicles have charge rates of 30 kilowatts. That means that if they are plugged in and charging for a whole hour, they will only be able to take 30-kilowatt hours whether they are charging on 50 kilowatts or 350-kilowatt charging equipment. Many newer models today have average realized charge rates of 60 to 80. While they will be able to get 60 to 80-kilowatt hours during a 1-hour charging session on 350-kilowatt equipment, they will only be able to get roughly 50-kilowatt hours on all the 50-kilowatt equipment. The average charge rate on EVgo's network today is 43 kilowatts the average charge rate on high-powered EVgo's 350-kilowatt equipment is in the low 50s. We forecast charge rates on our network to go up to the low 80s by the end of the decade, driven mostly by vehicle mix on U.S. roads. Utilization is defined as percent of time a vehicle is plugged in and charging on a 24/7 basis. We reported network-wide utilization of over 19% for December 2023. As I mentioned earlier, our underwriting models do not see utilization in excess of the mid-20s. Daily throughput per stall is calculated using the following formula: utilization multiplied by charge rate multiplied by 24 hours. We previously reported average daily throughput per stall of 201-kilowatt hours for December 2023. Average daily throughput is forecasted to increase considerably and reach 450-kilowatt hour in the current years, mostly driven by increasing charge rate. Now let us look into the individual stall cost structure. Each stall has 3 major cost categories, throughput defendant costs, total dependent costs and sustaining G&A. Let us unpack each of this category. Throughput depend on costs, as discussed earlier, today, a part of charging network cost of goods sold. We previously said that throughput dependent costs are roughly 60% of charging network cost of sales as of Q4 2023. We reported $18.5 million of charge in network cost of goods sold in Q4, which implies approximately $11.1 million of throughput dependent costs for the quarter. Dividing this number by reported throughput of 50 gigawatt hours, we get to $0.22 per kilowatt hour of throughput dependent costs for the quarter. It is important to note that throughput defendant costs vary over the course of the year. They are low in winter and high in the summer due to summer energy tariffs, which are generally higher than winter energy tariff. Our estimate of throughput dependent costs for the full year of 2023 is therefore $0.25 per kilowatt hour, and that is the number you see in the table on Page 5. We do not expect this number to change real dollars over time. In other words, we do not assume anything but inflationary prices to drive this number up. We previously said that stall dependent costs are roughly 40% of charging network cost of sales. We reported $18.5 million of charging network cost of sales in Q4 and which implies approximately $7.4 million of stall dependent costs for the quarter. The average number of EVgo owned and operated stalls in Q4 was 2,794, taking the average of quarter end Q3 and quarter end Q4, dividing $7.4 million of estimated stall dependent costs by $2,794 gets us to $2,600 per stall per quarter or $10,600 on an annualized basis. We have assumed that this number will improve by around 10% over the next few years, driven by operational efficiency initiatives aimed at reducing the maintenance cost per stall. We previously reported adjusted G&A of $27.2 million for Q4 and estimated that roughly 30% of this cost was sustaining G&A costs or $8.2 million. We also estimate that roughly 85% of this cost relates to our owned and operated network, while the remaining 15% relate to costs needed to sustain eXtend stalls and the flagship business. This brings us to $6.9 million of total sustaining G&A associated with the owned and operated network, dividing this number by $2,794. The average number of stall in Q4 would get to $2,500 per stall on a quarterly basis or $9,900 per stall on an annual basis. We forecast this number to significantly improve over the next few years, mostly driven by increased scale of the network, creating a leverage effect within the cost category. As we discussed during our Q4 call, our installed base is profitable on a stand-alone basis, which is seen in our illustrated one stall example showing average stall performance on an annualized basis. Average daily throughput per stall in Q4 was 192-kilowatt hours. We have also disclosed that roughly 15% of our network in Q4 had daily throughput of 350-kilowatt hours and better with the average being roughly 450-kilowatt hours per stall in this cohort. As you can see in the second column on Page 5, the economics of the top 15% of our stalls in operation today generate very attractive cash flows. As you can see, an increase of throughput per stall of 2.3x yields an increase of average per stall cash flow of 19.3x due to the leverage effect. We have also laid out the economics of the average stall we target to achieve in our network. As we have mentioned, we have assumed average throughput per stall will increase to 450-kilowatt hours driven mostly by increased charge rates together with a conservative utilization assumptions. And the leading edge of our network already demonstrates these levels of throughput per stall today. We have assumed no major changes in average revenue per kilowatt hour or throughput dependent costs, both expressed in real 2023 dollars. Modest improvements to stall dependent costs driven by a more efficient approach to maintenance and improvements to sustaining G&A, driven mostly by increased scale. Applying all of these assumptions, we forecast an average stall in a few years, generating 23x the cash flow versus our stall today. Our current deployment plan is for 800 to 900 new EVgo-owned stalls expected to be put in operation in 2024 with an average vintage CapEx per stall of $160,000 before offset. After netting out the approximately 40% of offset that we described on previous earnings calls, net vintage CapEx per stall is approximately $96,000. We often use terms vintage of fiscal CapEx, and I would like to clarify both of those. Vintage CapEx refers to total CapEx spent in the current and prior periods, which relates to charges than when operational within a specific time period. For example, 2024 vintage stalls and 2024 vintage CapEx refer to all stalls that will go operational within the year 2024. And all CapEx that is associated with these stalls independently of when it goes incurred. Fiscal CapEx refers to dollars spent in a particular time period independently of when these assets go operational. For example, 2024 fiscal CapEx will likely contain dollars spend on closing out 2023 vintage assets, 2024 vintage assets, 2025 vintage assets and standard miles for future developments for 2026 and beyond. In order to demonstrate further we have added an illustrative example of a stall that goes operation in May 2024. Vintage CapEx associated with such a stall is $160,000, as shown on the slide. The full number of $160,000 will be included in 2024 total vintage CapEx but it will be spent and thus included in fiscal CapEx in periods prior to May 2024. When we look at the period in which full vintage CapEx of $160,000 associated with this illustrated stall is incurred, we are looking at about $10,000 incurred in 2022 and thus included in 2022 fiscal CapEx, $75,000 incurred in 2023, and that's included in 2023 fiscal CapEx and another $75,000 incurred in 2024 and thus included in 2024 fiscal CapEx. Let us also go through capital offsets again. 40% of our vintage CapEx is expected to be covered by such offsets, which is a combination of OEM payments, grants and incentives. As we have discussed before, we receive approximately $33,000 per stall built under our partnership with GM. This is typically received within a couple of months of stalls going operational. We have over a decade of experience in successfully identifying, applying for and securing brand at the federal, state and local levels. More than half of our sites that go operation in 2024 have some form of a brand incentives attached to them. We usually collect those within several months after the site goals operational. In January this year, the IRS clarified rules about 30C eligibility essentially resulting in more sites being eligible for funding that we had previously expected. As a result, we now expect around 50% of our network plans to be eligible to receive 30C funding. We expect to monetize these credits by sales of a portion of our tax credit. We expect these transactions to be infrequent, 1 or 2 times a year, thus the lag between when and as it goes operational and monetization of the corresponding tax credit could be anywhere between 6 to 18 months. With that, operator, we can now turn the call over to questions.

[Operator Instructions] Your first question comes from the line of Chris Dendrinos from RBC Capital Markets.

Really appreciate you guys putting this together. It's super helpful. I guess maybe just starting off here on the tax credit. I mean should we be anticipating any type of sales this year? And then maybe just how should we think about pricing or valuation on those as maybe a discount to the total value of the tax liability?

It's Olga here. So we should expect a sale of our 2023 portfolio this year. I cannot give a more precise answer in regards to which quarter, but it will be this year and hopefully in the next few months. On a discount, we have some indications of it, we're still working through that and finalizing it. So I'll probably give an update to the market once we're closed to the transaction just to make sure that we've given the most up-to-date information. But we are definitely looking at a few opportunities right now to be able to transact.

Got it. And then I guess maybe as a follow-up, you guys have that deal with GM to get those payments for the stalls. Is there any additional conversations kind of going on? And maybe what's the appetite from the OEMs out there to continue funding for the network build out?

Yes. We talked to, obviously, GM all the time. We have a great relationship with them, Chris, as you know. And we're in dialogue with them all the time around the type of stalls that we put in the ground in terms of the customer experience in terms of the performance of the network. In terms of conversations with other OEMs, we're always in dialogue with other OEMs in terms of either charging credit programs or software integration and in some cases, there is conversation around other -- their support for infrastructure funding. Clearly, being able to deploy infrastructure relieves charging anxiety, which is still remains one of the barriers for getting people to kind of -- to make the decision to buy electric vehicles. So we think it's in everyone's interest for the charging infrastructure to get funded and installed and built.

Your next question comes from the line of Stephen Gengaro from Stifel.

So 2 questions for me, and I just wanted to go back, and I can't see on my screen anymore, the example that Olga laid out. But I think in the sort of target example, you're talking about $35,000 in cash flow per charger per year. Am I reading that right?

That's correct.

$37,000, correct.

Okay. I just wanted to make sure I saw that correctly. And then the 2 other questions. One is...

Well, what we're saying to -- sorry, Steve, in the middle column is what we're actually generating on an [indiscernible] basis for the top 15% of our charges today.

Okay. Great. I understand that. And then when we think about your -- either your installed base or as you sort of think about how your network is evolving. I believe most of what you're putting in right now is at the high end of the range, right, the 350-plus charges. But when we think about your installed base, is there any color you can give us kind of what it looks like now and how you think about it in like 2025?

Yes. So very clearly, Chris, we -- Stephen, sorry, we've got about 3,000 stalls in operation, as we said, as you know, about 1/3 are in the sort of 50-kilowatt range. and about 1/3 of the network is in that 350-kilowatt and then the other 1/3 is in between. And that's changed dramatically. So in our Q4 call, if you go back to the materials a few weeks ago, we said we had 425 350-kilowatt stalls in January '23. By the end of the year, we had over 1,000, so 1017, so we more than doubled the number of 350-kilowatt stalls. So it's gone from roughly 20% of our network in January '23 to about 1/3 of our network, 350-kilowatt stalls. Pretty much everything we're deploying -- so we said in our last call, we're deploying 800 to 900 OEM upgraded stalls this year, we had 930 last year. They're pretty much all 350-kilowatt tools. So the average speed of our network is clearly rising over time, which has a pretty significant impact on charge rate, obviously.

Okay. Great. Great. And then just one final one, and I know I bolt this probably during the earnings call. But when we think about your site planning and your track record, there's obviously very good and the algorithms that AI used to sort of determine locations and number of stalls, et cetera. How do I think about the Tesla impact and them opening your network and how that either impacts sort of your existing assets in the ground and/or your site planning, and I assume you're taking that into account.

Absolutely, we do. And so I think several things here. First, as their network gets opened up, we think that stimulates demand fundamentally for battery electric vehicles. It removes that charger anxiety issue that's on the minds of some folks. Secondly, we think of this as a completely different network, Stephen. Their stalls have historically been built around corridors or highways over many years built by Tesla to encourage people to buy electric vehicles and take the long-distance trips. If they're not in highways or corridors, they tend to be these mega retail sites with 20-plus stalls. That is a completely different network from ours. Ours are typically in urban, suburban kind of locations, high traffic places where you and I are going to kind of go grocery shopping or retail shopping, whatever it might be. We don't tend to go to our stalls to charge our vehicles. We're going shopping and in the parking lot of the Kroger's or the Target is one of our stalls. So they're a completely different network. 75% of our stalls don't even have a Tesla site in the same ZIP code. So again, if you think about going to -- going about shopping or getting a haircut or a restaurant, it's unlikely that a customer is going to go out of their way for a 20- or 30-minute charging station to a different ZIP code, which could be a 20- or 30-minute round trip just to get that experience when they're going to the Kroger's to get some milk or eggs or whatever. So it's a very different experience. It's a very different location. We have been deploying these chargers to accommodate all electric vehicles for, as you know, for a very long time. Tesla is beginning to open that up, and I think they're beginning to see the issues required to charge multiple vehicle models. So for instance, they've got shorter cable lengths. So if a non-Tesla vehicle, a CCS vehicle is parked in their stalls, they'll often end up taking 2 stalls instead of 1. There's plenty of YouTube videos that you can see to see how that can take place. The magic dock sites that are currently open that will charge CCS as well as Tesla vehicles have -- you could see that the plug scores, which is -- plug score is just like a yelp for charging stations typically have much lower plug scores. And so you can see that there's a challenge in terms of connecting the software between their charger and non-Tesla vehicles that they will still need to work through. So it's a different user experience. It's one that we think is going to result in either some cost or frustration or congestion of Tesla locations, which is, again, reinforces what we think, which is that our locations will see increased demand. And just to give you an illustration of how different the network is, our sites typically have 1.6x the amenities than a Tesla site that's within walking distance or within 1,000 feet at the stall. So I think that gives you a sense of how different the networks really are.

Your next question comes from the line of Craig Irwin from ROTH MKM.

This is very helpful. We really appreciate this. So looking at the illustrative cash flows of your stall performance, right, on a single stall basis. The top 15%, can you comment -- is this basically all 350s versus a network average in the fourth quarter that was materially below that? Is that a large part of the delta on the throughput? Or is it more geography and other factors that are driving a major chunk of this difference?

Yes. So it's a combination of factors. We definitely wouldn't claim that all 350s have higher throughput per stall in all 50s. That's not the case. Specific location plays a much bigger role. If you look at those top 15%, why is the high utilization site? Again, combination of factors. Most of them have higher share of rideshares, those are located in more downtown kind of higher traffic areas with the rideshare, drivers prefer to use them. They allocated in high utilization markets, not all of them, but most of them, you'll likely find them in a high utilization market that is Los Angeles, San Francisco, Houston and a few others. And some of them skewed towards being located in high MDG density areas. Again, not all of them because it's a combination of some of the right kind of prone sites will be not in the high density areas. It's a combination of things. What we would like to also emphasize here is that when we look at distribution of throughput per stall across different stalls in our network, and we look at historical distribution a year ago, 2 years ago, we see a very similar shape worth to be noted. So look normal shape of the distribution where you would always have a portion of sets which outperforms the average and you have average and you have laggers. What we're observing over the last 2 years that the whole distribution is shifting to the right. So what we expect will be happening in a few years is a continuation of that trend where the whole distribution will be shifting to the right and your top 15% of the sites will be exhibiting higher throughput per stall number than we've seen today, your average stall will be shifting to higher and your laggers also, by definition, will be siting as a function of the small vehicles and more traffic on our network. So -- and with that, we would always have a portion for various reason, performing better than average and average performance better than the laggers. Today, we also have among those laggards kind of like making a point about the new versus old, who is the laggers in our network today. It's a combination of a couple of things as well. It's either very, very new stalls, which literally just open and usually require like up to 6 months to get to the projected utilization levels or its stalls which are underperforming because they have very old equipment, which is waiting for to be touched upon our new program -- within our renew program or this is a stall, which also hold mainly equipment is performing fine, but the decision to secure these locations was made in prehistoric times, like 2016 and 2017 before we had those sophisticated network algorithms. We are constantly and we've talked multiple times throughout this type of cost were constant reviewing that segment of our network and making decisions either to upgrade equipment or close down equipment. So that kind of laggards segment of our network should continue -- or should expect to always have it as you always will have new site waiting for the utilization to come. But the portion which is associated with old nonperforming equipment, you should expect that to shrink in the next few months.

Okay. Excellent. So then your throughput in your target scenario that you lay out are 450, which is what you're achieving today on your top 15% of the network. So over the next number of years, equipment is going to continue to improve. And the nice thing is with the diversity of vendors you have, the quality and capabilities of the equipment that you buy, will change and evolve and improve for the experience of EV drivers. Is there a specific reason that you keep this throughput assumption flat? You did mention the renewal program. Obviously, that's something that's lifting the legacy base. But is there maybe room for the throughputs to go up? I assume commercial will be a bigger piece going forward and that could potentially lift that number quite dramatically?

Yes. That's absolutely right. The number you've seen is our target average number for the network in the next few years. When we are looking beyond that in the next, let's say, 3 or 4 years, when we're looking beyond that, we're definitely assuming that the average throughput per stall on an average stall, right, target net for average will go even further than what we are showing. Now if we're saying that 450 is a target average factor was saying at that leading edge of the network is definitely going to be higher than that, driven by exactly some of the trends I've been talking about and some of the trends you are mentioning now.

Craig, can I just add just so that we are clear, we are seeing increasing charge rates in our network. Charge rate is a function of the speed of the charger and the batteries in the cars. The assumption of going from 46% to 80% charge rate assumes no improvement in battery technology than we already have today. So in the vehicles that are being sold today, the 70-plus battery electric vehicles today, we've got something like the 80% of them are over 50-kilowatt charge speed, that's in the 10% to 80% state of charge more than half or over 90 kilowatt. We don't assume any improvements in battery technology. All we assume is that people are buying electric vehicles with the same technology today over the next several years. And so the mix of vehicles on the roads just increases over time. And secondly, we assume that we continue to deploy 350-kilowatt chargers. We are -- that's what we're doing today. It's what we did last year. And so the mix of chargers on our network also improves over time, and that's what results in the increase in charge rate. And as Olga said before, we assume a relatively more conservative utilization of the network that's consistent with our business case in the low to mid-20s. And that's how you get to 450, which we think, is a conservative number, as you pointed out.

Excellent. Excellent. And then last question is really a point of clarification. So your target network average that you share with us, is this something that we're maybe considering in a single-digit or even low single-digit number of years? Or is this something out over the next decade?

It's probably something like 3 to 5 years.

Your next question comes from the line of Gabe Daoud from TD Cowen.

Thanks for putting this together. A lot of great detail and color, incredibly helpful. Could we maybe just go back to the -- about the last question and Olga, I guess you mentioned it's maybe a 3- to 5-year target. What are what are maybe some of the assumptions that you have embedded in here in terms of the size of the EV fleet over the next couple of years in order to sustain a 23% utilization rate?

Yes. So we -- Gabe, we're assuming that we're deploying at the 800 to 900 stalls a year this year. That's -- you can assume that's the run rate that we're doing right now. And I think as you...

Okay. From a stalls standpoint, that 800 to 900 a year?

That's right. That's right. I think as most people forecast that the rate of growth of battery electric vehicles. And again, when we talk about this right now, most of the vehicles that are charging in our network are non-Tesla vehicles, which have currently experiencing a higher rate of growth than Tesla vehicles for new sales. but the rate of growth of battery electric vehicles exceeds and is expected to exceed the rate of growth of chargers. That's something that I don't think anyone is seeing a change in their expectations.

That's a good point. That's helpful. And then I guess just as a follow-up also, maybe just sticking to some targets over the 3- to 5-year period. If you do achieve this target network average across the network. What does that -- or how does that translate to what like the adjusted gross margin looks like at the corporate level over the next like 3 to 5 years? How does that number expand over time?

So it will expand. We probably would hold different events where we'll talk you through that. But you also can do the modeling and divide it by yourself, right? Like if you look at the elements we've just depicted on the slide, throughput dependent costs and stall dependent costs. They sit in the cost of goods stall, sustain engine per stall sits in adjusted G&A. So by like applying your own math and running different sensitivities if it's all fits, if it's the type, if it's little lower, you yourself could see what the adjusted -- implied adjusted gross margin will be.

Yes. No. No. That's right, Olga. And we see a good number. I just wanted to if you can maybe dial that in to be a bit better. But no doubt, that's helpful.

So it's 40% of our cost of sales is fixed on a personal basis, which we're showing here, and we disclosed last time. And as you could have seen last time the margin on the charging business has significantly expanded year-over-year just because of the leverage effect that sits in cost of sales.

Your next question comes from the line of Bill Peterson from JPMorgan.

I wanted to actually talk more about the market. I think you talked about one of the first slides is around the network throughput gigawatts in us coming in ahead of your expectations. I think the spec deck actually had closer to 250 around this time frame. So I'm not sure what time frame we're talking about. But in any case, I think that the DOE data says you have around 7% share of fast charging, tests are up 55%, you're looking at 800 to 900 stalls per year. So I'm trying to get a sense of how do you see the share of the market today of this network throughput. And with that install rate that you're talking about, how do you see the share of market progressing?

Yes. So look, I think it's important to think about the market in terms of gigawatt hours as opposed to stalls, at least that's how we think about it because we make money...

That is my question, what is your share of the market in gigawatt hours? How should that track when you add stalls?

Yes. Well, I think it depends on what everyone else is doing as I think most of us can see that other than Tesla, there's not a lot of growth in stalls outside of, I guess, ourselves and Tesla. That's been the case last year. And so I think you have a point of view on the growth of other companies, but we expect to see the 800 to 900 stalls growth. We think that's very supportive in terms of maintaining the utilization of NAV or growing the utilization. And we'd love to increase the rate of growth. We have the growth engine that we think has the capacity to increase and to be scaled up and that's just a function of the capital that we'll need to do that.

And what do you expect your share of the market was with this 200 gigawatt hours?

We don't -- I don't -- we don't have a -- we don't really forecast a market share that we can share with you. It would depend upon what we think others are doing in the market.

And it will depend what percent is DCFC will end up being. So there's a bunch of factors that go into, but we're just not disclosing that.

Okay. That's fine. So just coming back to this target network, the cash flow reaching, I guess, what, $37,000. For the sustaining G&A per stall, and I think this target, you said 3 to 5 years out, what -- would you probably back into it, but what is the expectation of G&A growth over that time frame?

What is the expectation of the growth in total G&A?

Yes. It's -- to achieve the sort of target of 3 to 5 years, the sustaining G&A per stall comes down, but presumably it goes up again, we could probably back into it. I'd just like to quickly understand how you sort of think about G&A...

I think I understand your question. So the sustaining G&A that I think I mentioned in my remarks, will definitely scale with more stall, but the sustain engine per stall will go down as a function of some efficiencies and some fixed elements baked into it. Just to disclose what this particular number is target -- is actually conservative because it's calculated with a target of 7,000 stalls. So it's a little more than doubling the size of our network versus today. When we go beyond that, we should expect that number decreasing in further, but not as much as versus -- like doubling versus it. If you double it again, you probably will see some efficiencies but not as much as government for the first time. Does that answer your question?

Yes. No, that's helpful.

Your next question comes from the line of Stephen Gengaro from Stifel.

Two quick ones. One, I maybe overthinking this a little bit when you think about utilization of the assets and you think about sort of the evolution in EV buyers, right? Because my sense is, right now, you're buying these -- a lot of times, you can charge at home. Are you taking into account like the likelihood that as EV adoption kind of accelerates into places where people can't necessarily charge at home that there's a utilization impact?

Yes, for sure. I mean that's what we've seen. We've seen, as we've said before, 6% utilization 2 years ago in December 2021 to 8% to 19% by December '23, and that's been driven by more VIO. It's been driven by vehicle miles traveled. People driving electric vehicles are driving almost the same number of miles as an ICE vehicle today. But it's also been driven by significant growth in DC fast charging, the share of DC fast charging. So folks who live in multiunit housing who don't have a private driveway are able to access these stalls more conveniently. They're faster today, our network. And so it's a shorter time to be able to charge the vehicle whether they're out shopping. We're also seeing significant growth in rideshare. We've talked about that going up from 11% to 25% in the last 2.5 years as a portion of kilowatt hours in our network. And that's folks who are using DC fast charging. So yes, we think that, that is a pretty significant driver of the increased utilization, which is therefore a driver of throughput per stall per day.

Great. And then just one quick follow-up to one of the things that Olga mentioned earlier was when you talk about credits in California, is that a meaningful portion of revenue? I mean can you give us any color on sort of the impact that has on the top line?

Yes. So we're talking about 2 types of credit here. The great in California regulatory credits, which you can see we report in and on they're not as meaningful they don't have as much of a meaningful contribution to our revenue as a few years ago, but it's a separate reporting line, which you can very easily extract. We're also talking about the tax credits and tax credits related into 30C, which is a new -- it's an IL deflation making to the center. Those will not be part of the top line once we sell-off our portfolio that's just going to hit cash flow but not P&L.

Your next question comes from the line of Doug Becker from Capital One.

I just want to reiterate, I think this is very helpful. In the examples you were using the average CapEx of $160,000, which I guess is the expected average for this year. What's the expectation going forward in that 3- to 5-year period? And specifically thinking about the prefabrication approach, and you just opened up a site in Texas. Does that actually allow that number to go down over time? Or do you other inflationary pressures keep it around that $160,000 level?

Yes. No, it does, Doug. So you're right, the $160,000 is for this year. That's before the offsets. As a reminder, the 40% of offsets that we expect. But look, we are -- I would say that we are in the very early stages of this industry. There's what, 40,000 -- 30,000, 40,000 DC fast chargers installed across the United States and most people think that by 2030, that will be, I don't know, 250,000 to 350,000. So I look at this, and I think there's a -- we're 1/10 of the way there over the next rest of this decade, maybe a little more than that. That's a tremendous amount of growth. And so I would expect us and others to see the CapEx per stall reduce over time. We are -- it is clearly an area of focus for us. We have not shared or disclose where we think that will get to. But clearly, it's an area that we're focused on and the prefab skids that you mentioned is just the first of some examples that we're quite focused on.

And just any high-level thoughts on how many sites are candidates for the [ prefed skids ]?

We haven't shared that, Doug. So -- but it will be a reasonable number, but that's -- again, that's just the first of a set of initiatives that we hope to start talking about over the coming -- over the rest of this year.

Your next question comes from the line of Chris Dendrinos from RBC Capital Markets.

You talked a bit about how you guys differentiate versus Tesla. And I think one thing with them is they are willing to go into a market pretty early to hopefully drive demand, I guess, in that market for their vehicles, whereas I think you guys are more driven by the profitability side of things. So when you enter a market, how should we think about kind of that initial utilization rate? Is there a minimum that you all kind of target before you go into a market? And then is it, I guess, a 3- to 5-year time line to get to that network utilization target of 23%? Is that how we should think about that?

Yes. So the answer is when we evaluate and enter in the new market, we're always looking at the overall NPV that the network in that market. will generate for us. And we are underwriting to 10 years, as I mentioned earlier, the 10-year unlevered pretax IRR. And how do you get to that? You can get there by a variety of different means. You can build on that and already expecting that utilization is going to be high, aka, 20% plus in the first 6 months that there are locations like that. You can get there by building in a very low energy cost utility territory or you think we just don't need as much utilization to get to that return or you have so many -- it's such like a really grant environment that a lot of your upfront investment get offset right away in asset 40% off that we're talking about is average, right? Some of the stalls will be higher than that and some of the sales are lower than that and so for some of the locations where you really have a lot of offsets, you also don't need as much utilization. So there is not a single number we have and might say, well, okay, everything is above these utilization in the next 6 months, we'll build an everything below, we're not building a noose thinking about it. We're thinking about it on the 10-year perspective. And for some geographies despite the fact that we're not going into the geography too early where the site will be idle for years before cars come in. We do think about building a little bit ahead, and we're okay to have an average utilization for the first couple of years, knowing that this is a geography where year 3, for example, a little bit burst on EV adoption and then the remaining 8 years, where we will show high utilization and get our 10 year back. So it's all about optimization for the whole network on IRR basis versus just simple utilization targets.

And we have no further questions in our queue at this time. I'll now turn the call back to Badar Khan for closing remarks.

Great. Well look, in summary, EVgo is on a mission to deliver a fast charging experience that leaves fossil fuels in the rearview mirror. Electric vehicle adoption and the EV charging industry is seeing unprecedented growth. And as you've heard, we are building our business utilizing our proprietary network planning and charger underwriting to locate the best chargers while delivering double-digit IRRs. Olga and I look forward to sharing our progress with you on our Q1 call in just in the coming weeks. Thanks very much, everybody.

This concludes today's conference call. Thank you for your participation, and you may now disconnect.