Before we begin, please note that today's discussion contains forward-looking statements on the environment as we currently see it. It does involve risks and uncertainties. Our filings with the SEC, including our most recent earnings press release and annual report on Form 10-K, provide more information on the specific risk factors that could cause actual results to differ materially. Some of the numbers we will reference in our discussion today are non-GAAP. You will find additional information on our non-GAAP financial measures, including reconciliations, where appropriate, to the corresponding GAAP financial measures in our most recent earnings release and other filings with the SEC. Good morning. I'm John Pitzer, Corporate Vice President and Head of Investor Relations. It's my pleasure to welcome everyone to the next installment of our Investor Webinar series.
The purpose of this forum is to give our outside owners the opportunity to engage more closely with our leadership team and focus on a specific topic that we believe is critical to our strategy, helping our owners better understand how we intend to drive long-term value for all of our stakeholders. In January, we hosted our first webinar on the PC Client business, where we highlighted our view of the long-term market and the strategies we are pursuing to capture, share, and value. In March, we provided a deep dive into our data center and AI business, where we addressed the breadth of our product offering, including the launch of Emerald Rapids in the second half of 2023, Sierra Forest in the first half of 2024, and Granite Rapids shortly thereafter.
We also discussed our strategy to truly democratize AI from the cloud through the network to the enterprise, client, and edge, and from data prep to training to inference with our suite of silicon and software assets. Today, we turn our attention to our internal foundry model, which will create a foundry-like relationship between our manufacturing groups and our internal product business units. To drive transparency for our owners, as of our Q1 2024 earnings, our manufacturing groups, including internal manufacturing, technology development, and Intel Foundry Services, will be combined into a single reportable segment on par with CCG, DCAI, NEX, and Mobileye. I would note, today's webinar will focus on the operational transformation we are undertaking with our internal foundry model.
While we will touch on Intel Foundry Services as a part of today's discussion, we intend to host a much more detailed and exhaustive webinar on IFS in the second half of this year. Joining me today are David Zinsner, Intel Executive Vice President and Chief Financial Officer, and Jason Grebe, Corporate Vice President and General Manager of Intel's Corporate Planning Group. Dave will provide some historical context on the evolution from IDM 1.0 to IDM 2.0, which has given rise to our internal foundry model, and frame the financial benefits of this model and how it will transform how we run the company. Jason will speak to specific examples that will enable the transformation and be a key driver of our committed $8 billion-$10 billion in savings and efficiency gains exiting 2025.
He will also explain how this operational change will be a tailwind to our IFS strategy. After their prepared comments, Dave and Jason will be happy to address questions. I would remind you that we will report Q2 earnings later in July and would ask that each of you limit your questions to the subjects addressed today. Each participant will be able to ask one question and a quick follow-up. With that, let me turn it over to Dave.
Thank you, John. It's good to be here with you today to provide an update on our internal foundry model. Based on conversations that we've had with shareholders over the last several quarters, we concluded there was an opportunity to provide more clarity around our internal foundry model that underpins our ongoing transformation from IDM 1.0 to IDM 2.0. Pat's first focus when rejoining the company was to regain process and product leadership. With our five nodes and 4-year strategy proceeding well, positioning us to regain transistor and power performance leadership by 2025 and strengthen our client roadmap and significant improvements in our data center roadmap, we're now focused on accelerating efforts to drive best-in-class cost structure, which is what the internal foundry model is intended to facilitate.
We first discussed our internal foundry model in Q3 of 2022 as part of our multi-year cost efficiency effort, which includes reducing costs by $3 billion in 2023 and $8 billion-$10 billion exiting 2025. As discussed in Q1 earnings, we're on our way to achieving $3 billion in savings this year, including a $2 billion reduction in OpEx and $1 billion reduction in cost of sales. We'll provide another update when we report Q2 earnings. The internal foundry model is a key enabler of the incremental savings to achieve our $8 billion-$10 billion we previously committed exiting 2025 and is critical to our longer-term ambition of 60% gross margins and 40% operating margins. In our current model, we have P&Ls for Intel Foundry Services , or IFS, and the product we use of CCG, DCAI, and NEX.
The manufacturing and technology development groups allocate 100% of their costs out to these P&Ls. As we transition to the internal foundry model, the manufacturing group will be accountable to a standalone P&L for the first time. The reportable P&Ls will now be the manufacturing group, inclusive of IFS, manufacturing, and technology development, our product groups of CCG, DCAI, and NEX, and all other. Under this new structure, we'll move from a cost allocation to a market-based wafer test and packaging price from our manufacturing groups to our external and internal customers for both front-end fab manufacturing and back-end test and packaging services. This is a shift from today's practice, where cost is allocated from the manufacturing group to the business units. Pricing will be based on comparable industry benchmarks from foundries and OSATs based on performance, power, area, and cost.
This will provide an economic relationship akin to a fabless and foundry partner, shift the association from today's support model to a customer-supplier relationship. It's important to note that this is not only a significant change for our manufacturing groups, it will also unencumber the business units from a large portion of their allocated costs, allowing them to predictably optimize their own OpEx and investment decisions. To put a finer point on this today, our manufacturing, technology development, and IFS groups comprise roughly 40% of our headcount, 25% of our OpEx, and more than 90% of our CapEx. Our ability to benchmark and track the effectiveness of these investments has historically been obscured by the lack of transparency inherent in an allocated cost model. Our pivot to giving manufacturing its own P&L is critical to identifying cost reductions.
This new model will add transparency and comparability that exposes the true economics of the business by more directly measuring the financial performance of our teams with peers. It will also drive accountability as we harden the connection between decisions and costs. For example, the financial impact for decisions around capacity allocation shifts to manufacturing, while the impact of design attributes and extra stepping or late demand changes will directly impact the business units making those decisions. We're already seeing benefits of this newfound transparency and accountability and expect to see many more as teams are incentivized to drive better financial performance. This model isn't a panacea to our cost challenges. It's simply a structure that enables us to drive accountability and measure progress. We still need to identify and drive the changes necessary to improve margins.
Jason will delve into several of these opportunities, but first, I'd like to spend a few minutes explaining how we got here, the rationale for the change, and the operational impact we think it will have. Since our founding in 1968, we have been an Integrated Device Manufacturer or IDM, a company that both designs and manufactures its own semiconductor chips. Intel was tremendously successful executing this strategy, what we call IDM 1.0, for many decades. We had two guiding principles for manufacturing under IDM 1.0. First, always maintain capacity buffer, given our strong market position, and second, move to the next node as quickly as possible, aligned with Moore's Law. Additional capacity was key, as the cost of ceding market share outweighed the incremental capital cost.
Unfortunately, this also had the unintended consequence that the BUs could drive excessive expedited wafers and frequent changes to wafer loadings without economic consequence or accountability. We had a rapid ramp up and ramp down approximately every 2 years, where cost reduction was largely achieved by moving to the next node versus optimizing costs on existing nodes. We had leadership and process technology of up to 2 years versus competition that led to consistent gross margins in the 5a%-60% range. This was the right strategy for the time, and we were highly successful. The semiconductor industry and computing in general has evolved rapidly, and we need to adjust our business operations in response. With the advent of smartphones, the digital network, the intelligent edge, cloud computing, and AI, compute demands have diversified.
In the latter half of the last decade, we competed in diverse end markets with a strategy that was highly tuned to win in a single end market. At the same time, the diversification of workloads, devices, and product lines led to development of a vibrant foundry ecosystem, shifting from a niche to the norm. Furthermore, the industry is experiencing a significant increase in capital intensity as cutting-edge nodes grow more expensive. Finally, there is an ongoing technology pivot to disaggregation that will drive longer tails, as some IP will remain on older process technologies that doesn't benefit or require the shrink that would come from the next node. This increases the criticality of cost optimizing existing nodes versus simply moving all capacity to the next node.
These changing industry dynamics, coupled with our own execution missteps that caused us to lose process leadership, have contributed to the shift to IDM 2.0. On one hand, this is a significant change for the organization as we adapt to new market dynamics. On the other hand, our objective remains unchanged: to maximize value for our shareholders. While we transition to this new model, we want to be clear about the inherent value of the tight connection across our manufacturing, business units, and IFS. We're better together. The IDM model provides four key advantages. First, it enables our technology development, manufacturing, organization, and business units to co-develop the process for custom product needs. Second, it enables us to get products to the market faster.
Intel's internal CPUs are the ramp vehicle for Intel 18A, which will come to market in 2025 and enable us to regain process leadership. Third, the model gives us a substantial tailwind as an external foundry by ramping our process nodes on internal volume and de-risking the process for external customers. Lastly, under IDM 2.0, we'll have an increased scale and a broader ecosystem of EDA, IP, and design services that will further strengthen our IDM capabilities. To clarify the financial structure, let's review a mock P&L that demonstrates how the new model will operate. In the IDM 1.0 structure, manufacturing organizations do not accrue any revenue and only incur cost of sales and operating expenses. Both cost and spending are then fully allocated out to the partner business units, ultimately leaving the manufacturing groups with no gross or operating margins.
As such, the sum of all Business Unit P&Ls equals the IDM P&L. As we transition to the IDM 2.0, the manufacturing group will have a standalone P&L. Revenue will be based on wafer sales to the Business Units at market pricing, while costs will continue to be based on true manufacturing costs. As such, the manufacturing business will begin to generate a margin. OpEx will no longer be allocated out, but will remain on the manufacturing P&L, leading to a manufacturing operating margin. The price-based manufacturing revenue will now drive Business Unit cost of sales. Therefore, Business Unit margins will be based on market-based cost structure. The intercompany sale of wafers from the manufacturing P&L to Business Unit P&Ls will be eliminated, causing the Intel-level P&L to be consistent between IDM 1.0 and 2.0 in this hypothetical example.
Our long-term ambitions are to achieve non-GAAP, 60% gross margins and 40% operating margins. The internal foundry model will largely help us to achieve this in the following three ways. First, having a full P&L will enable clean benchmarks to third parties and highlight opportunities. This changes incentives for leadership, which will further lead to an optimized cost structure for both the manufacturing group and business units. We've highlighted before that our ambitions are to be the second-largest external foundry by 2030. That remains our goal. In the new model, based off of internal volume, we expect to be the second-largest foundry next year, with manufacturing revenue of greater than $20 billion. Second, pivoting from allocated cost to price will change behaviors within the business units.
For example, it'll encourage business units to move to the next node faster, as they won't be burdened by the initial high cost as a new process ramps. It will also incentivize test time optimization, fewer expedited wafers, and more selective use of samples as the business units see the direct impact of their choices hit their P&L. Third, the manufacturing group will now face the same market dynamics as their foundry counterparts. They'll need to compete for volume through performance and price, as internal customers will have the option to leverage third-party foundries. To attract external foundry volume, they must do the same. This will lead to more effective adoption of standard third-party practices. Over time, our internal business units will have increasing flexibility to choose where their products are manufactured. This is not a new concept for Intel.
Today, roughly 20%-25% of our silicon is manufactured externally, and Meteor Lake, which will launch in the second half of this year, benefits from both internal and external wafer supply. While increased flexibility is the plan of record, with five nodes in four years proceeding well and expected leadership on Intel 18A, in conjunction with the new accountability driven through the internal foundry model, we're confident that our manufacturing group will continue to be the partner of choice for our internal business units. In fact, we continue to be confident that we will sign up our first Intel 18A external foundry customer this year. Now, let's pivot to the financials and opportunity in front of us. Intel's gross margins compare unfavorably to industry benchmarks.
Leading fabless and foundry competitors, combined into a virtual IDM, would achieve roughly 70% gross margins, which is 20 to 30 points higher than Intel's recent results. We took a big step forward this year as part of our efforts to reduce spending. As we've worked to implement the strategic changes required for our internal foundry model over the last several months, we've already identified multiple cost savings opportunities. While I'm very optimistic about our future and long-term model, it'll take time before our manufacturing group P&L is world-class. As you can see, we initially expect a negative operating margin %, but the significant opportunities stemming from our internal foundry model will begin to drive accretion to Intel.
I'll now hand it over to Jason Grebe, Corporate VP and General Manager of Corporate Planning, to address specific examples of how our internal foundry model will reduce costs and drive efficiencies for Intel.
Thank you, Dave. It's great to be here with you and the investor community. As Dave mentioned, in IDM 1.0, our business units would make decisions based on product cost provided by the internal manufacturing teams. In a cost-based model, the business unit P&Ls would absorb changes in cost, ramp impacts, and allocated expenses that can vary over time. In IDM 2.0, as we grow Intel Foundry Services, we will leverage our at-scale factory network, while also providing market-based pricing for our products and services to external customers. The internal foundry model combines the best of both worlds. We will extend the use of market-based pricing to our internal business units, offering them the same stability as our external customers. The internal foundry model combines the best of both worlds.
We will extend the use of our market-based pricing to our internal business units, offering the same stability as our external customers. We maintain the intimacy and deep connection between our product groups and technology development teams, preserving the competitive advantage we had as an IDM. Finally, we provide a tailwind to our external foundry business by effectively creating the industry's second-largest foundry, letting external customers build off our internal scale. For example, we will have more than five internal products being developed on our latest process technology, Intel 18A. Our internal products and the early learnings around them essentially de-risk the process for our external customers. As a team, we've already done a lot of internal analysis and benchmarking to identify areas of opportunity. Let me talk through a few examples we've uncovered and where we've already started to see the benefits of our internal foundry model.
Let's start with expedites. Expedites are requests to the factory to accelerate production or sample materials. While expedites provide faster routes for specific lots of material, they disrupt the efficiency of the factory network and lower the overall output, which leads to incremental capital and spending. Our benchmarking suggests that we expedite materials around 2x to 3x more than our industry peers, with an estimated 8%-10% hit in overall output. In the internal foundry model, business units will be charged 1.5x to 3x the price of a wafer for expedited services, just like a customer would at an external foundry. This will tie the financial implication of an expedite request directly back to the business unit that requested it, giving the decision-maker a clear understanding of the return on investment of that decision.
This transparency, combined with greater operational controls, will drive additional accountability and reduce overall expedite requests while retaining customer service levels. As the number of expedite requests come down, the manufacturing group will be able to optimize their factory flows, and we estimate annual savings in the range of $500 million-$1 billion over time. Next up is test and sort cost. As we shift to the internal foundry model, business units will now be charged a market price based on test time. Historically, Intel has enjoyed a structural cost advantage on its test platforms. This is driven by the high level of parallel testing we have spent years developing. However, due to this capability, we've increasingly grown our test times compared to our industry peers. Currently, we estimate our test times are 2x to 3 x those of our competitors.
Effectively, our lower-cost testing platform was subsidizing the growth in our test times. By switching to the internal foundry model, we are uncovering design choices and trade-offs that have impacted test times. As we sharpen our focus, we are addressing test strategies within design and in partnership with our quality network to reduce overall costs while still preserving the quality that Intel prides itself on. We anticipate this focus to save around $500 million annually over time. As you can see, we've identified many opportunities for optimization across both our business units and our manufacturing organizations that will lead to significant savings. In the interest of time, I'll briefly review a few other activities we're currently exploring. Product business units will leverage stable pricing, which will help accelerate our ramp rates, leading to better capital utilization and faster yield curves.
We estimate this will save up to $1 billion. The internal foundry model also has benefits beyond the efficiency of our manufacturing network. We are already looking at several areas within our product design environment to improve efficiency there. For example, we expect to see increased design efficiency as we start to charge industry rates for tool utilization, just like our competitors, which has hundreds of millions of dollars of cost savings opportunity. We also have plans to minimize samples, reduce steppings, and economically target our advanced packaging, all areas with massive cost savings opportunities in the range of $500 million-$1 billion annually. We are also implementing industry-standard systems and tools to be able to scale up our revenue without increasing operational costs. We estimate that internal foundry model will help drive closure on our overall $8 billion-$10 billion savings goal.
Beyond the cost and efficiency benefits I've mentioned, the internal foundry model is a significant tailwind for our Intel Foundry business as well. Everyone on this call knows that it is critical for our foundry business to be successful. We heard directly from our customers about what they need from us. Number 1, a portfolio of process technology and IPs. Number 2, supply assurance. Number 3, protection of their data and IP. Number 4, world-class service levels. We know that IDM 2.0, with our commitment to deliver five nodes in four years, will address the portfolio of process technologies that our customers expect, while also improving our IP portfolio. The internal foundry model addresses the next three needs, starting with supply assurance. Intel Foundry Services will have the authority to allocate corridors of capacity to its customers.
Overall, by establishing the manufacturing organization as its own business and ensuring it has the decision rights to manage its P&L, we expect to be able to communicate clear capacity corridors and supply commitments to our external customers. The next concern we hear is related to protection of customer data and IP. The internal foundry model is an arm's-length approach that increases the independence of our manufacturing organizations. We will deliver complete segregation for foundry customers' data and IP. As we begin retooling the company for this transformation, we are architecting with a security-first mindset, taking data separation as a key tenant in our systems design. Our customers expect world-class foundry service levels. As we've already begun the internal foundry model shift, we are building the service-oriented mindset that is absolutely required to be a key player in the foundry business.
The manufacturing group, including our Intel Foundry Services organization, are benchmarking themselves against industry peers to ensure we're on track to provide best-in-class service levels expected from a foundry. Now, let me briefly summarize what we talked about today. Evolving from IDM 1.0 to IDM 2.0 has been a critical step for Intel to win. On top of that, establishing an internal foundry model is one of the most significant steps we are taking to deliver IDM 2.0. Our internal foundry model is the mechanism that makes IDM 2.0 a success by fundamentally changing the way the company operates and establishing the structure and incentives needed to change our culture and drive new behaviors. By leveraging industry standard planning processes, data management strategies, systems, and tools, we are building the foundation to be a world-class IDM and a foundry provider.
Giving the manufacturing group their own P&L, along with the associated transparency and accountability, will be a key driver in achieving our target to reduce $8 billion-$10 billion in cost and achieve our long-term profit goals. Finally, the internal foundry model will be a strong tailwind to our IFS strategy.
Thanks, Dave and Jason. We're now going to transition to the Q&A portion of the webinar. Jason is going to be joining me live in the studio. We have Dave joining us today virtually for Q&A. As a reminder, we will ask each of you to ask one question and a brief follow-up question, where applicable. Jonathan, can we please take the first question?
Certainly. One moment. Our first question comes from the line of Ross Seymore from Deutsche Bank. Your question, please.
Hi, guys. Thanks for letting me ask a question. Thanks for all these great details. I guess for my first question on the manufacturing side, you have the slide, didn't have a scale on it, but you talked about the operating margin being negative initially and then climbing over time. Can you just give us an idea of, is that mainly cost dependent, the $4 billion-$5 billion that Jason highlighted, or is there a significant revenue dependency within that? If it is revenue dependent, is that more internal or foundry IFS based?
Dave, why don't you go ahead and start, and then Jason can follow up?
Yeah, good question, Ross. Obviously, you know, obviously, scale is going to matter and, you know, level of revenue is going to matter. I would say, you know, as we look out over the next few years, you know, most of our revenue is going to come from internal foundry customers, so that's really where we're expecting most of the lift from. You know, there's a significant amount of improvement that we think we can make, and most of the remaining $8 billion-$10 billion after we get through the 2023 portion, which is $3 billion, you know, will be in cost of sales and will be directly benefiting, the, you know, the foundry, P&L. Those will be a key aspects.
You know, obviously, also, we look at our OpEx as a percent of revenue in the foundry business, and it's relatively high, given, you know, our need to accelerate, you know, our no transitions and the investments necessary to do that. You know, obviously, we'll get back onto a normalized cadence. My expectation is that OpEx starts to, you know, improve as a percent of revenue as we progress, and that should provide a good tailwind to improving operating margins as well.
Perfect. Ross, do you have a follow-up question?
Yeah. I forgive me for going a little off script on this, but I noticed this morning you talked about or announced a sale of a portion of your IMS business. Can you just remind us the strategic importance of that business and why you sold a portion of it today?
Yeah, sure. We'll try to stay on the, I, the foundry, internal foundry topic, but just since it was newsworthy today, I think I can touch on it. You know, really delighted that we were able to close a sale of a portion of the IMS business for an overall valuation of $4.3 billion. This will turn out to be one of the best acquisitions we've ever made given that level of valuation and the investment we made at the beginning. You know, this business is really in a good spot right now, given its multi-beam mask writing tool set, which, you know, is, you know, essentially directed at EUV and ultimately, High-NA EUV .
You know, when Bain came in, they saw the upside to the business going forward. We sold about 20% of the business to Bain. I think Bain is just a terrific partner for this, and I think really reinforces how valuable this business is and the opportunities that are in front of it. You know, by the way, we think this is another kind of example of our focus on unlocking shareholder value. As you know, we did our follow-on offering for Mobileye. We've done 2 offerings for Mobileye that have unlocked value for shareholders and, you know, created a valuation for the overall business.
This will be another opportunity to do that and get some proceeds by selling about 20%, but also establish the value of that business as well. You know, terrific transaction. It's been in the works for some time. We think, you know, giving it a little bit more independence, having Bain involved, who, you know, just has, I think, a lot of, you know, provides a lot of strategic value working with the team there to drive the business, you know, will ultimately yield a terrific result for the company and ultimately also for Intel. You know, maybe just since, you know, we're talking newsworthy items, maybe I'll just hit on another one. There's been a couple of announcements of late on expansions or greenfield sites around the world.
Of course, the numbers are big, which, you know, will be spent over time. Obviously, they come with incentives, many of which are front-end loaded, so pretty beneficial to us and all within the idea of, you know, getting the cost aligned with, you know, our competitors and also, you know, helping on the cash flow side. Nothing that we announced is any different than what we gave as our view at the Investor Day and how we've talked about it over the course of the quarter since then. We still believe, you know, the investment phase of Intel, which is kind of the 2022, 2023, 2024 time frame, you know, we'll roughly be spending it at a net CapEx intensity in the 30s as a % of revenue.
We still feel, like, confident that we will be able to modify that thereafter, into a level that's somewhere in the mid-20s as a percent of revenue. That, combined with, you know, getting back to process leadership, getting back to product leadership, in markets that are, you know, growing at pretty healthy rates, we think, you know, should give us very good free cash flow. We still feel committed to the roughly 20% of revenue coming in the form of free cash flow.
Ross, one thing I might add on IMS, it's a nice reminder that we've been investing as a company in the EUV ecosystem for well over a decade. As Pat mentioned on the last earnings call, with the ramp of Intel 4 and Meteor Lake, we're rapidly closing sort of the gap on EUV, and we think that the IP and kind of industry perspective that IMS gives us becomes even more important as we transition to High-NA in the back half of the decade. Jonathan, can we have the next question, please?
Certainly. One moment for our next question. Our next question comes from the line of Vivek Arya from Bank of America. Your question, please.
Thanks for taking my question and for this informational webinar. Let me paint a hypothetical scenario. You know, let's say one of your product groups wants to go to an external foundry. You know, what flexibility do they really have? Because these foundry relationships are done years in advance, and even if that product group leaves, who covers that underutilization charge, right? Doesn't in some way this new structure kind of give the illusion of separation and flexibility without actually having the real flexibility that your fabless and foundry competitors are enjoying today?
Jason, maybe you should start on this one, and then Dave, follow up.
Yeah, sure. I mean, over time, our business units will have increasingly more flexible decisions about where they manufacture and where they drive their IPs. We actually have our x86 IPs at external foundries today, that won't be a major change for some of the business units. They do have the ability to flex out where their products require it. Reminder that we are investing in five nodes in four years to catch up from a process technology perspective. We feel good about that investment and the progress to date. The manufacturing company, on their side of the house, will financially incentivize our product groups to want to manufacture on those improving process technologies internally. We expect the majority of the volume to remain in-house. The business units for years have had the flexibility to flex out for product needs where they want it.
We're actually having products in 2024 begin to start that process, so it's not a major change for us, and it provides them the flexibility and the architectural choices necessary for them to have the best competitive roadmaps from a product perspective. We feel comfortable with it.
Dave, anything you might want to add?
That was a good answer, Jason. Maybe that's the only other thing I would just say, which you pretty much touched on, is, you know, that this because we're measuring external or internal foundry as a P&L or foundry as a P&L, you know, they're highly incentivized now to grow their P&L, which means they're highly incentivized to grow their revenue, to load their fabs. You know, they're gonna bend over backwards to do that. And so, yeah, you know, there'll be pressure because there'll be competition as we progress over time. That just makes... You know, competition makes people better, you know, makes companies better.
You know, my expectation is that the foundry business rises to the challenge, delivers process technology, delivers service levels, delivers capacity in a way that, you know, excites our customers, whether they be external customers or they're internal customers, and they bring volume because of that.
Vivek, do you have a follow-up question?
Yeah. Thank you, John. Why shouldn't we think of this as kind of Intel's first step towards potentially breaking up into a fabless and a foundry business? What would be kind of the pros and cons of such a move?
Dave, do you want to start?
I think, yeah, I'll start, although Jason, I think, hit on a lot of it in the prepared discussions. You know, we think there's a ton of benefit for having both a business, product business, and a manufacturing business combined. You know, we, I think it enables better process technology over time. It enables better products, because you're working closely with the TD groups on a regular basis.
We think that, you know, our ability to use our internal customers as what we call customer zero, you know, to ramp volume on the new nodes is a benefit to the external foundry customers, who then get to benefit from the use of those nodes that have, you know, have been worked out, you know, many of the early kinks to the process. You know, we think that we'll be able to leverage the EDA partners in a better way, leverage IP in a better way because we have this combined entity. We don't see this as a requirement to, you know, to split the business. We actually think quite the opposite. By doing this, it actually helps bring the businesses together in a more wholesome way.
You know, the other thing we're gonna do, you know, we have seen, you know, some commentary out there about how external customers will view this, and we see this as a tremendous positive for external customers. You know, they do want, you know, as Jason said, they want protection of IP. They want to know their corridor is going to be there for them. By, you know, creating this separation, but having it under the same umbrella, they get that confidence that we will do that. In fact, we will have two legal entities. We'll have two ERP systems to manage things so that we can, you know, do all the things that our external customers are expecting of us as we kind of roll out the foundry model to them. Jason, anything to add?
No, I think that was a good answer, and I would just add, you know, improving the accountability, the transparency, and the decision-making straight back to their P&Ls is gonna make both organizations better. We view it as a better together story.
Perfect. Thank you, Vivek. Jonathan, can we have the next question, please?
Certainly. One moment for our next question. Our next question comes from the line of Aaron Rakers from Wells Fargo. Your question, please.
Yeah, thank you, for doing this conference call and all the details. I guess I just wanted to ask about as kind of you guys mentioned in the webinar, you know, this diversification and kind of process towards chiplet architecture and process node diversification. How do we think about the Tower, you know, acquisition and how that fits into the strategy as we think about, you know, the evolution around, you know, leading edge versus mature nodes and so on and so forth in the IFS strategy?
Dave, you want to start with that?
Yeah, I'll start. You know what? I think the way I view it is there's kind of three legs to the stool of being in the foundry business. There's participating at the leading edge, which obviously this helps tremendously in. There is, you know, then transitioning what was a leading edge to, you know, more a lagging edge or legacy node, which we intend to do over time and, you know, generate better ROI over time for our nodes because we're extending the likes of those nodes. And then there are, you know, kind of more specialized offerings, you know, that are provided, you know, by foundry suppliers to, you know, to foundry customers.
That's really where we see the Tower business being so important for us. You know, it'll be one of the three legs of our offerings is, you know, just being able to provide the, you know, the more specialized process technologies. You know, from a P&L perspective, they'll have their own P&L, obviously, that we will track and measure internally. Externally, you know, it'll consolidate up into, you know, the foundry business that we are going to measure and talk about externally beginning next year in the first quarter.
You know, it'll be part of what that group will be driving in terms of improved performance, trying to drive more gross margin and operating margin leverage from every one of those three stools over time.
Aaron, do you have a quick follow-up?
Yeah. I do. I wanted to ask about the expedite or I think sometimes referred to as the hot lot dynamic, you know, in the, in the manufacturing process. You mentioned an 8-10% output, you know, impact that you've seen. I'm just curious, you know, as you've, you know, already started to go down this path on the IFS strategy, you know, how has that already evolved? How has that changed within the operational, you know, rigor of the company? Is that 8-10% already coming down? Any data points that you can share around that impact to the model?
Yeah, maybe I'll take that one.
Please do.
What we've seen over time is that both on the engineering side, or early development, as well as the high volume manufacturing side, we drive a lot of expedites into the factory, trying to pull in material and expedite material through, which at the end of the day, really slows down the factory efficiency and overall tools, tool set implementations that we're driving. What we started to do, both from a process perspective and from an accountability and traceability perspective, is attack that problem right away. We've kind of operationalized process flows within Intel to start to minimize those, prioritize those, and get those out of the system wherever possible.
We've already started that work immediately. At the same time, what we're doing is now being able to track those requests and push back the bill for those back to the business units, so they understand not only that there's an impact on manufacturing, but if they were doing that in an external foundry per se, there would be a significant P&L impact to them for the choices that they make around expediting material. In 2023, before we stand up the P&Ls, we're actually starting to surface those expenses to the business units to say, like, "Hey, if we continue to operate the same way that we typically would, this would be the type of impact that you would see on your P&L as we go forward." We're starting to minimize those things on manufacturing.
In the long run, what we hope to do is be able to drive out overall cost, because we can take down capital, because we can run our tool sets more efficiently in manufacturing.
Aaron, thank you. Jonathan.
Very helpful.
Can we have the next question, please, Jonathan?
Certainly. One moment for our next question. Our next question comes from the line of Mark Lipacis from Jefferies. Your question, please.
Hi, thanks for taking my question and for the great presentation. Really appreciate it. So the first question, does this new strategy require you to retool existing fabs or change how you build fabs going forward? The reason I ask is my understanding of one of the reasons Intel had been so successful in the past in staying at the leading edge, is that you were a focused IDM with products focused on x86 processors, so you could, you know, design and build factories focused with that product set. My question is there extra CapEx to retool your existing factories or higher CapEx intensity for fabs going forward as you bring foundry customers online?
Jason, why don't you go ahead?
I would say there's not an incremental CapEx bill to bringing on incremental foundries, incremental customers online for the Intel Foundry Services business. We have to deliver PDKs to them, so they can develop on top of our process development work that we're doing. From a CapEx perspective, there's no significant CapEx increase for them other than the fact if we have incremental volume, obviously we'd have to have a bigger footprint of overall capacity. Again, this is one of the advantages that we have, is we have a global network of capacity with space already in our plans, so it would just be us building out according to the demand that we get from our new foundry services customers.
You know, Mark, I might add, in the IDM 1.0 model, the cost driver was really maintaining that 2.5-year process lead at almost any cost, and it evolved into things like Copy Exactly! and fast ramps up and fast ramps down of nodes. In the new reality of IDM 2.0, we're going to look for many other levers to try to drive those cost reductions by trying to run our manufacturing footprint just a lot more efficiently than we did in that IDM 1.0 model. Do you have a follow-up, Mark?
Yeah, that's helpful. I did have a follow-up, if I may. You know, I think I have and others have viewed Intel as kind of venture capitalists for Moore's Law or the process research. I think you guys were the first to bring high-k metal gate and strained silicon and FinFETs to market. My question is, where do those investment dollars hit your financial statements now? Whose P&L were they in, where will they land going forward? Who has responsibility for that? Thank you.
Dave, do you want to start with that?
Yeah. Most of the ones you referenced, you know, maybe like a future one, might be backside power or what we call PowerVia, that we've invested in. Those will still show up when you look at the consolidated P&L as research and development expense. When you look at it by business unit, it will show up in R&D for the, whatever we end up calling the foundry business unit, as R&D. It will, you know, somewhat be kind of incorporated, because, you know, the foundry business has to get a return on all these investments, you know, through their wafer pricing. It will show up in the BU P&Ls ultimately as, you know, part of the wafer pricing, that they pay, for, you know, to the foundry business unit.
I'm sure there will be instances where it's, you know, where, you know, the more research-oriented spend is more of a product investment as opposed to a process investment. In those cases, it would show up, again, on a consolidated space, it wouldn't show up in R&D, but in those cases it would show up as R&D in one of the business units or, you know, allocated across several of the product business units, and likely not show up in R&D for, you know, for the foundry business unit. It's a little bit complicated because there are some exceptions. There are some things that, you know, will be investments that are done at the corporate level, that Pat will kind of manage on his own.
In those cases, you know, we'll allocate it broadly across all business units if it's one of those kind of projects. Assuming it's something more specific to either the product area or, the foundry or, you know, manufacturing and TD area, it'll show up in those respective business unit P&Ls.
Mark, thanks for the great question. Jonathan, can we have the next question, please?
Certainly. One moment for our next question. Our next question comes from the line of Gus Richard from Northland. Your question, please.
Yes, thank you very much for taking the questions. The first one I have is, you know, as you go on this journey, you're looking at the external foundries and benchmarking yourself against them, and I was wondering how you were doing, particularly as it pertains to tape-in, as I understand, takes a lot longer at Intel, and also cycle time within factory, because t he way I understand you on your factory, you have limited number of processes, limited number of products, and it just makes more sense from a cost perspective in that model to have a longer cycle time than what is typically observed in a foundry. Any help there would be appreciated.
Jason, great question for you.
Maybe I can start. On the foundry side, obviously, we have to be competitive from a tape-in perspective, we have to be competitive from a throughput time perspective, and we plan to be world-class there. Our customers aren't gonna give us a break there from a foundry services perspective, so we're absolutely attacking those problems. We understand exactly where the foundries are relative to us in that space, and we're attacking those problems, and our plan is to be world-class and competitive there. From an internal perspective, obviously, the business units are driving the products that they want to be able to go drive, and the factory units need to be able to support those business units and support those tape-ins. That intimacy that we've had for, you know, 50+ years will remain.
We wanna become world-class there, and we think by separating out the P&Ls and letting them drive the behaviors independently, will get us to a better answer collectively at the company level.
Gus, do you have a follow-up question?
I do. Just, you know, historically, you guys have had a very high level of equipment reuse. You know, are you know, for each node, gonna build a new factory with a fresh set of equipment and then just, you know, no more hand-me-downs, or just reuse the older equipment for older nodes?
Yeah, maybe I can start. It's going to be case by case, depending upon the demand set that we have and the factory network and tool sets in those factories. Obviously, reuse does save us some capital and drive some efficiency. We'll obviously drive to that wherever we can, but in the grand scheme of things, it's just going to be a question of what demand sets we have. We are planning for longer tails on our process technologies to try to get more return on those investments, but we'll continue to monitor it as we go.
Dave, anything you'd like to add on that?
No, other than to say, you know, not every node has been identified as a great external foundry node. Obviously, Intel 16 is, Intel 3 is, but the nodes between those two, we've determined aren't, you know, what we want for foundry offerings. In those nodes, we'll have, you know, still high reuse of equipment as we progress into, you know, more advanced nodes. Those that are, you know, clearly targeted as external foundry nodes, you know, clearly we'll attempt to extend the lives and get better ROC. It is going to drive, in some cases, more greenfield expansion.
That's part of why you're seeing more greenfield expansion from us, and the announcements that, you know, were just came out about, you know, one of them being a greenfield, is in part because of that. You know, we'll be trying to extend the lives of some of these fabs, and, you know, as everyone knows, when you can, you know, take these beyond their depreciable lives, they generate really great cash flow and have really accretive growth margins to the business. That is our goal, yeah, to drive a better P&L ultimately for shareholders.
Gus, thanks for the questions. Jonathan, can we have the next question, please?
Certainly. One moment for our next question. Our next question comes from the line of Ambrish Srivastava from BMO. Your question, please.
Hi. Thank you. Thank you very much for hosting this call. Dave, I want to come back to the external versus internal. You said earlier on, which is correct, you said Intel has been having external foundries, but the whole purpose behind this is to make the P&L more transparent to the product groups. Do you envision a scenario where the external goes from 20%, 25% to 30%, 40%? If so, what is the implication for how do you modulate your CapEx capital intensity if the pendulum was to swing in that direction?
I mean, obviously we would adjust capital intensity if that were to be the case. I would just say that I think that's unlikely. I think that, you know, as we progress the foundry business, you know, which will be a combination of internal and external, is gonna execute really well, you know, based on all indications on how process technology is going with respect to the five nodes in four years. You know, I think that, you know, based on the notion that, you know, performance will be good on those nodes, and pricing will be very competitive for those nodes, I think, you know, the business unit, product business units are gonna see a lot of financial incentive to wanna go with the internal, the internal foundry.
You know, that said, there'll be situations where that doesn't align with timing or needs, and there's always the flexibility that they have, which they've had for all this time to utilize external foundries. You know, I think percentages we're throwing around are probably what will be the likely percent of external foundry that we have within the business, you know, when we look out 5, 10 years from now.
Ambrish, do you have a follow-up?
Yeah, I had a quick one. On the savings, potential savings... Sorry. Thanks for that, Dave. The potential savings of $4 billion-$5 billion, what is the timing that you guys expect to realize those savings?
There's two, I think there's two answers to that. When Jason showed all these ideas we had and how much they rolled up to, some of those ideas, you know, we think we can get in the next year or two. Some of them are longer term. That said, you know, like I said, we, you know, we have this target of $8 billion-$10 billion of savings, exiting 2025. $3 billion of it, we're targeting to get completed this year in 2023. That would, you know, leave roughly $5 billion-$7 billion of cost savings, still to go, that we expect to get, you know, as we exit 2025.
Some of that will be the things that Jason highlighted, in terms of, you know, benefits of this internal foundry model. Some of it will be, you know, quite honestly, blocking and tackling type spending reductions that we'll implement, in terms of, you know, just driving more efficiencies. We're doing better in terms of, you know, ship cost and getting the best cost across everything we buy and those kind of things that will also contribute to the remaining $5 billion-$7 billion.
Perfect. Thank you, Ambrish. Jonathan, can we have the next question, please?
Certainly. One moment for our next question. Our next question comes from the line of Vijay Rakesh from Mizuho. Your question please.
Yeah, hi, and thanks for doing this event. Just a quick question, just a clarification first. When are you looking at foundry revenues going to $20 billion next year? When do you expect to start breaking out the internal versus the foundry business? I have follow-up. Thanks.
We, we already provide right now the foundry, you know, external foundry revenue base, and that, you know, in our current construct of segment reporting. Obviously, we expect to evolve that next year in the first quarter. You know, I don't know exactly how we'll provide it, but our expectation is that we will still provide insights as to how much of the revenue coming from our foundry business unit is coming from external sources versus that is coming from, you know, internal product groups.
Vijay, do you have a follow-up?
Got it. Yeah, and thanks, John. I mean, if you look at foundry, it's really a scale business. Is there a risk that when the initial volumes are subscale, that, you know, it might be a headwind to the margins, you know, versus, you know, some of the other foundries that could be running at scale? You know, and likewise, you could have some foundries that are pretty good at certain nodes that could be driving a lot of volume and be very competitive and might be.
Mm-hmm.
- headwind to your process. Just want to think through that. Thanks.
We're obviously subscale. That's why our operating margins are, you know, where they are, what we highlighted as profit, where they are below zero. We do have to build out the scale of that organization. Obviously, we're, you know, trending below normal trend lines for the overall business. The expectation is that recovers, you know, that improves, you know, our rate of volume for the foundry business, you know, aggregate foundry business.
That said, I think, you know, as Pat correctly concluded as he came into the company, you know, because this is a scale business, because of the CapEx intensity, you know, more volume is required through those fabs than what can be provided through the Intel product groups you know, that's why it's so important that we move into, the foundry business to attract more volume, to drive more scale, to make it more profitable, and obviously, you know, to allow ourselves to generate more cash flow that we can, you know, in a lot of ways, reinvest back in, to advance the process technology and advance the product offerings.
Vijay-
That's-
Sorry, I was going to add, Vijay, that, you know, the idea of talking about $20 billion of foundry revenue in Q1 of next year is to show you that we are an at-scale manufacturer foundry. That would make us the second largest foundry in the world. Now, all of that business is coming from internal fabless customers, not external. As we add those external customers, it just increases that scale. Jonathan, I think we have time for one last question, please.
Certainly. One moment for our final question. Our final question comes from the line of Srini Pajjuri from Raymond James. Your question, please.
Thank you. Hi, Dave. You sounded pretty confident about announcing a 18A customer, external customer, later this year. Just trying to understand, when you say announcing a customer, what exactly that entails. I mean, are you anticipating, you know, signing a volume agreement? Just if you could give some color on that. What are some of the issues that you still need to resolve before you announce a customer?
Yeah. Okay, well, let me take the first ask, part of that question, or sorry, second part of the question first. you know, I think the main still to-go item, in terms of, securing a customer is just the maturity of the PDKs. you know, I think they're getting to where they need to be, but they're not quite there. I think until they are at the right maturity level, you know, there'll be a lot of activity, with, prospective customers. Having one really sign up for some volume, I don't think that happens until we have, you know, the PDKs at a mature level.
You know, we're expecting that in the back half of the year, and I think that's why we feel, you know, based on early read from customers, I think that's why we feel confident we'll be securing a customer in the back half of the year. You know, obviously, we're talking about 18A volumes, which, you know, won't be produced until, you know, 2025 and, you know, and beyond. There won't be any incremental revenue you see from an 18A customer, you know, this year or next year, even early the following year. Our guess would be it's either 2025 or beyond that we start, you know, having volume from those customers. You know, we'll have some, you know, formal agreement.
did it with, you know, the other two customers we've secured, one on Intel 16, one on Intel 3. We generally have some, you know, some sense of what the volume requirements are or what the aggregate dollar amount is that we're talking about, what SKUs, they're being directed at, you know, those kind of formal items. We'll have to see. Obviously, you know, we'll have to see how things play out. But, in all likelihood, we'd be looking for some sort of prepays as well, in addition, so that, you know, that we can use that working capital improvement to accelerate our investments in 18A.
Srini, do you have a quick follow-on?
Yeah, just a more of a longer term. Dave, you kind of talked about the synthetic GM between the foundry and the fabless customers being close to 70%, and your long-term target is 60%. I would say, you know, if you get to 60%, a lot of us on this call will be, you know, very happy, given where you are today. Just curious as to, is there any structural reason as to why you should be below, given, you know, some of the advantages that you mentioned of being an IDM?
Yeah, I'd say, it's about the way you said it. Let's get to 60, then we'll worry about how to get to 70. You know, we think we have a good path to 60. You know, there are some, you know, at the, when you combine, one of the fabless companies, that operates, you know, more at the leading edge with, you know, the best-in-class foundry supplier, you know, That particular foundry supplier has a lot of scale, and there is a benefit, just a strict benefit to scale more than anything. You know, that's obviously going to impact the relative gross margins. That said, you know, there are probably other things that, you know, will be more beneficial for us, that we can also take advantage of.
You know, we'll have to see how things go. I think for now, I want to drive the team to 60, and then once we get to 60, we can think about what the next goal should look like.
Perfect. Thank you, Srini. With that, we've come to the end of this webinar. I'd like to thank everyone for all the great questions and the engagement this morning. We look forward to interacting with you all at the end of July as we report our Q2 earnings. Before we do sign off, though, I would love to hear your feedback on today's event and where you'd like us to focus on future webinars. Please take a couple of minutes out of your time to do a brief survey that you'll see on the web link. Thanks again. We'll be talking to you soon. Appreciate it.