February 17, 2022
Rearchitecting Intel for Growth – Pat Gelsinger
• Four takeaways from today. Intel is rebuilding their “Grovian” execution engine, they have the right strategy, they are leveraging their cores strengths to both grow in traditional markets and disrupt new ones, and “Intel is the next great growth story.”
• The sustained long-term demand for semiconductors are driven by the “four technology superpowers”: ubiquitous compute, cloud-to-edge infrastructure, pervasive connectivity, and artificial intelligence
• For the rest of the decade, they see leading edge growing at a 12% CAGR and trailing growing at 2%, leading to more than half total revenue coming from leading edge by 2030
• Pat reiterated their commitment to five nodes in four years, first set as a goal in July 2021. He said they are today ahead of schedule. One way they said they will do this is by partnering more closely with equipment vendors. Intel has always been paranoid about their IP and this closed off their relationships with the equipment companies, so this is a major change, if they are indeed acting differently
• Arrogance continues to be the first word that comes to mind when I listen to Gelsinger. It could be that he is just so proud and passionate about Intel and their rich leadership history, but I still find it off-putting. Intel fell behind TSMC and Samsung in manufacturing because of their arrogance. A couple times he referred to the money they are investing and their manufacturing sites as “his.” I would prefer he remember the money, fabs, etc. belong to shareholders.
• Software is an important part of their strategy. He also led with “open” being central to software. Again, openness is a major change for Intel.
• Intel revenue growth forecast: 2020-22: low single digits, 2023-24: mid to high single digit, 2025 and beyond: 10-12% YoY growth. That’s a bold statement
• They see one million units per day PC market to be sustainable, with low-single-digits annual growth. They believe Intel can grow their PC business by low to mid-single digits
• In the Data Center, Intel does a new product one year, then an upgrade the next year. Data Center growth will be mid-to-high single digits revenue growth up to 2025, increasing to mid-teens percent revenue growth in 2026 and beyond.
• Networking and edge will grow revenue at a mid-teens annual percent CAGR
• Traditional businesses – client, data center, networking – account for half of their future revenue. The other half will come from graphics (which they have not been leaders in previously – see NVIDIA passing Intel in market cap), Mobileye (“unquestioned leader in ADAS and AV”, still planning to IPO this business), IFS – Intel Foundry Services (they expect to announce major customer commitments later in 2022). Beginning with the April earnings release, they will provide updates on their progress by the new business unit structure.
• The strongest reasons to create an Intel Foundry Services is twofold. First, IFS gets all the process development spending for Intel’s products for “free”. Second, equipment utilization in fabs will be much higher since tools can be used in both fabs, for longer. These are both hugely valuable advantages that fewer than five companies in the world have.
• IFS capex will become meaningful in 2025 and 2026. Before that, Intel is leveraging their existing manufacturing capacity to develop and prove IFS to the marketplace.
• Gelsinger’s goal is a “double double”. He seeks to double Intel’s earning and to double the earnings multiple on their stock price. Hence, his goal is to 4x the stock. This would give the company a market cap of ~$800B. For comparison, NVIDIA’s market cap today is $535B and TSMC’s is $515B. This is a bold but not unreasonable goal, if Intel can defend their current territory in computing, grow their foundry business to be maybe 30% market share, and succeed on some other lines of business.
Data Center and AI – Sandra Rivera
• They will build their future strategy on Xeon Scalable processors
• Data Center and AI is estimated to be $30B today, growing mid-teens CAGR to $65B by 2026
• She and another person clumsily went through an on-stage demonstration intended to show a head-to-head between Sapphire Rapids and an NVIDIA processor. I don’t take this kind of theater seriously because there are so many ways to tune a computing “test” to show about any desired outcome.
• Sandra presented a lot of figures measuring how widespread Intel is in the data center segment. This isn’t very meaningful since Intel has 90%+ market share in data centers. Most important to Intel is defending this share from AMD, NVIDIA, and the in-house silicon of hyperscalers.
• AI is estimated to be more than a $40B silicon TAM by 2026, growing between 20% and 25%, depending on which part. The presentation didn’t specify if this silicon TAM is just processors or includes other types of chips, such as memory.
• Intel’s DCAI revenue is expected to grow mid-to-high single digits CAGR in 2021 to 2023, and then mid-teens CAGR from 2023 to 2026. They are basing the case for Intel to continue to lead in DCAI on their current market position and IDM business model, and to deliver cloud computing experiences
• Another meaningless head-to-head demo was included here
• Intel’s “tiling” approach to processor packaging allows Intel to more finely tune product design and performance to the specific sub-segment
Client Compute Group – Michelle Johnston and Jim Johnson
• The business has grown from $32B of revenue in 2015 to $41B today
• The COVID pandemic has transformed the way people use PCs. They (along with Microsoft) believe we have moved in the western world from one PC per home to one PC per person.
• CCG imperatives are to deliver leadership products and to deliver innovative experiences with Intel’s open ecosystem
• In their closing statements, they said “win back” market share in PCs is one of their primary goals
CFO Presentation – Dave Zinsner
• He has four objectives: revenue growth, gross margin expansion, healthy cash flow, and strong shareholder returns
• He said something telling early in the presentation; that he will instill financial discipline throughout Intel, implying that this is lacking today. From my experiences with Intel as a JV partner, this is true. Their two decades of large margins made them unfocused on cost competitiveness. Intel is financially sloppy and wasteful, from what I have seen. If they can improve this lack of a cost-focused culture – no small task – it will have a noticeable expansive effect on gross margin.
• The total market size for the combined Intel segments is estimated at $450B by 2026
• Their emerging businesses are headlined by IFS and Mobileye, but Intel is assuming strong growth in both emerging and traditional (DC and PC) to reach the double-digit revenue CAGR they are forecasting by 2026
• Intel bought Tower to learn how to succeed in the foundry business
• Gross margins will be in the 51% to 53% range in 2022 to 2024 and growing to 54% to 58% by 2025 and beyond. Lower foundry margins will be the drag on GM for the next few years as they invest in this business and scale it up.
• Dave included a walk between the 51-53% range and 54-58% in the future. The walk is: 1-2% from recovering process technology leadership, 1-2% from better ASP for their products, 1% from scale and product mix, and 1-2% from fiscal discipline. He says from what he has seen already, the 1-2% gain from fiscal discipline is a conservative goal.
• Their long-term capital intensity goal is ~25%. Capital intensity is capex as a percent of revenue. Part of the higher capex intensity in the next couple years is the investments they are making in fab shell space. This is building the relatively cheap fab building well ahead of need, so the expensive equipment can be added more quickly when demand dictates.
• Smart capital is: early shell investment, government incentives, pre-payments from customers for foundry services, partnering with Brookfield to share financing of some of their capex, and using third-party foundries. This last one is for Intel to use foundries rather than invest in nodes where their demand is too low for those investment to be cost-effective, or where the demand is uncertain enough that there is high risk of unused capacity.
• Their goal is to get free cash flow to 20% of revenue by 2026, among other things, Dave sees a lot of opportunity in working capital
• Capital allocation priorities: invest in the business (talent, technology, capacity), strategic M&A, return of cash (committed to a growing dividend). Dividends will be the primary way Intel returns cash to owners
• 2022 outlook: revenue of $76B, gross margin of 52%, $3.50 in EPS, slightly negative free cash flow
• Financial model in the investment phase during 2023 and 2024: revenue growth mid-to-high single digits, gross margin 51-53%, operating expense 28-30% of revenue, net capital intensity ~35%, neutral adjusted free cash flow (operating cash flow minus net capex and operating leases)
• Long term model: revenue growth of 10-12%, gross margin of 54-58%, operating expenses of 25-27%, net capital intensity of ~25%, adjusted free cash flow of ~20%
• The financial models Dave went through are their commitment – they are shooting for higher performance than what he described
Question and Answer – Pat Gelsinger and Dave Zinsner
• They don’t believe there is anything heroic in their assumptions about the size of the PC market and what Intel’s share of it will earn
• Responding to a question on whether or not gross margins will stay above 50%, both Pat and Dave said they will manage the business to that level of margin, with the multitude of levers they have within Intel to control costs and manage their product mix
• Pat said the data center space will be more of a dogfight than client. He believes Intel has regained the lead in the PC business and AMD is “in the rearview mirror” and they won’t lose the lead again. As arrogant as he comes across much of the time, this comment sounded more like confidence than arrogance.
• They are shooting internally for close to 20% ROIC
• The Intel turnaround story is a “trust me” story. This is the nature of any several-year plan in semiconductors. At the end, investors have to decide if they believe the management team or not.
• One analyst asked how Pat got so many good people back to Intel. He gave a compelling answer. He believes the people they have now is on a mission to turn around a company that is iconic both to Silicon Valley and to the United States.
• They are losing market share in Data Center in 2022
• The latency to see results from changes in strategy, such as entering the foundry business and new product development, are long enough that the company needs to be making several changes at the same time. The central example is turning their manufacturing base and process technology around, and at the same time developing new products that will use the new manufacturing processes, presuming the turnaround will succeed.
• Implied in their capital expenditures model is 10% of the total worth of offsets, between government incentives and partner contribution. Management feels this is conservative and he is targeting for this to be more than 20%.
Executive Breakout Sessions
Accelerated Computing and Graphics – Raja Koduri
• Goal is to reach $10B of revenue by 2026
• PC Era → Mobile + Cloud Era – now entering → Pervasive Intelligence Era
• Accelerated computing platform TAM will be $155B in 2026, $100B in silicon and $55B in software. Intel is differentiating themselves from NVIDIA with an open system, different from their competitor’s closed system. Intel’s goal of reaching $10B in this market is a 10x from where they are today but is less than 10% of the estimated total market at that time.
• Raja has the same condition Pat does, sprinkling adjectives throughout his presentation describing how amazing Intel’s products are. It feels over the top to me. I don’t know if it is confidence and excitement, or a continuation of the arrogance that caused Intel to lose their leadership in the industry. But he is a great speaker – strikingly good.
• The open software layer they are putting on top of their graphics hardware is called “Endgame”
• Falcon Shores is their next generation supercomputing product
• Celestial is their next generation product in Visual Compute. Intel splits accelerated computing into Visual Compute, Super Compute, and Custom Compute
Environmental, Social, and Governance – Cristy Pambianchi
• Intel has been a leader in environmental and sustainability, publishing their first such report in 1994
• The company’s 2030 goal includes using 100% renewable electricity and being net positive in water usage
• There are 121,000+ employees worldwide, 89% of which are in technical roles. They hired 17,000+ technical external hires 2021.
• She spent a segment of the time listing a group of technical leaders they got from competitors and other big names. I think this was included to try and show that Intel has reversed the brain drain.
• Leadership has reintroduced “Results” as a company value, to get back to the “Grovian” culture that was part of Intel becoming a leader in semiconductor manufacturing
IDM 2.0 – Manufacturing – Keyvan Esfarjani
• Key messages are investing to win with IDM 2.0, deploying smart capital, and transforming their operating model through IFS
• Intel has fabs in Oregon, Arizona, New Mexico, Israel, and Ireland. They have Assembly test facilities in Costa Rica, Chengdu, Vietnam, and Malaysia, and advanced packaging in Malaysia. I am not including Ohio and Germany here as these fabs were only recently announced and construction is just beginning. They are not operational.
• They have 35,000 employees in global manufacturing
• Manufacturing is ahead of schedule on their 4, 3, 20A and 18A nodes on their development and deployment
• Intel is moving from System on Chip to System on Package. This uses a group of smaller die in a single package rather than one large die. Assembly and test must be more advanced for this to work, but it brings advantages in die cost and development.
• The company will use both internal and external fab capacity for foundry needs, with both growing in absolute value through 2026
• Their assumptions on capital offsets (grants) in their IDM 2.0 investment plan are conservative
• IDM and IFS complementarity has a big effect on capital efficiency in front-end fabs. The sharing of process development costs and fab investments between these two businesses is, along with catching up on process technology, one of the two most important pillars of the turnaround strategy.
Intel Foundry Services – Randhir Thakur
• IFS has five key themes; a broad foundry portfolio, a large and growing market opportunity, a strong start, accelerated by the acquisition of Tower, IDM2.0 strength accelerates IFS path to profitable growth, and IDM2.0 and IFS complement each other
• IFS will offer process nodes from half micron down to leading edge. The tail of this is filled in by what comes into the company with Tower.
• Intel forecasts the foundry market, $95B in size today, will grow to $180B by 2030, with 30% of that future TAM on mature and specialty nodes and 70% on modern and leading-edge processes. 85% of this future $180B market will be automotive, mobile, and compute.
• The fab expansion in Arizona and the greenfield in Ohio are called out in one slide, suggesting this is where they IFS business will start manufacturing
• IFS will have a dedicated automotive group to focus on this fast-growing segment and its unique needs
• Randhir said the word “customer” many times during his presentation. Intel has a long history of dictating terms to their customers and other stakeholders, so this is in part to insure potential customers they will behave in a more collaborative and service-oriented way in the future, in my view.
Mobileye – Amnon Shashua
• Revenue was $1.4B in 2021, up steadily from $143M in 2013
• They think of the gap between ADAS and AV not as a range of capability, but as increasing the mean time between failures (MTBF)
• Mobileye sees two futures; robotaxis and consumer AV. They are the only company betting on both of these futures.
Network and Edge – Nick McKeown
• This is a new business unit, formed in 2021
• Key takeaways: networks are moving toward software, more computing is happening at the edge, NEX (this business unit) was created to deliver fully-programmable infrastructure, Intel has the right vision, relationships, and technology to lead in this industry
• The TAM for network and edge is $44B today, growing at a 12% total CAGR to $62B by 2026. Nick thinks of this in two pieces; “network fixed function” and “Edge and Network”. NFF is flat over the next five years. All the growth comes from Edge and Network, which they believe will grow at a 17% CAGR to 2026.
• Nick is at Intel because he has evangelized software eating as much of networking as possible. Pat offered him the chance to do that widely with all the markets that Intel touches.
Process Technology – Ann Kelleher
• They expect to reach performance per watt parity in 2025 and leadership in 2025. They are currently the leader in packaging.
• Intel 3 extensively uses EUV technology. Following Intel 3 is the first node in the Angstrom era, the Intel 20A node.
• Intel will be the first company in the industry to receive a high-NA EUV tool, which they will begin using in production in 2025
• Intel 7 is the current lead node. Intel 4 will be manufacturing ready in 2H22 and is the first node to use EUV. Intel 3 improves performance per watt by 18% over Intel 4. It will be manufacturing ready in the second half of 2023. Intel 20A is a 15% improvement in performance per watt and is the node. that introduces RibbonFET and PowerVia. Intel 20A will be manufacturing ready in 1H 2024. Intel 18A is a 10% improvement in performance per watt. This node will be manufacturing ready in 2H 2024.
• SH comment: with the difficulties that come with Moore’s Law bending, Intel’s advanced packaging and tiling strategy will help the company close the process competition gap with TSMC and Samsung
Software – Greg Lavender
• Intel is committed to a “vibrant open ecosystem for developers.” The opening of this section wants to convince analysts that Greg is the right leader for software at Intel, and that their software ecosystem is open and committed to the needs of developers.
• They use artificial intelligence to enhance their security software
• The presentation on this topic was all technical and product focused. There was no mention of the size of the market or the revenue the company hopes to bring in from software. I believe this is because software for Intel is part of the hardware they sell and not a separate product.
Executive Breakout Summary
The key messages from the breakouts are the complementary nature of the new foundry business and Intel’s existing manufacturing network and the focus on changing the culture of Intel back to one focused on results. Mobileye and Network is a lot of technical detail that certainly means a lot to subject matter experts in these areas. As an investor, I have to ask how much Intel will get out of these segments. I think Mobileye is a winner and will continue to win, because I think collected data and learning are large competitive advantages in this emerging area. In networking and edge, I see a bet on a combination of McKeown and Intel’s technology, scale, and relationships. The breakouts had a lot of technical detail which I interpret to be aimed at experts in the specific areas. The typical investor or analyst will get little out of that level of detail. Even for the technical details of the areas I do understand deeply, such as process technology, the thesis for investing in Intel still comes down to whether or not one believes they will execute on their plan. It all looks good in the slideware.
The headline from this has to be Gelsinger’s goal of a “double double” for the company. He seeks to double Intel’s earning and double the EPS multiple in their stock price. This is bold, targeting a 4x increase in the stock price of a company the size of Intel, but it is not impossible. The stock price is beaten down now and the opportunities they are taking on are large, especially foundry. The biggest reach among the goals they put out is the long-term revenue growth range of 10% to 12% annually. I will do a “what is Intel worth” post later. Without even doing that model I think if you believe Intel will grow revenue 10%+ annually after the turnaround the stock is a no-brainer buy today. Qualitatively, the keys from this investor day are fivefold. First, the way their new foundry business and their existing IDM business will complement each other is hugely valuable. The only other company in the world that has that advantage in capital efficiency is Samsung, who has it between their foundry and memory businesses. For Intel the advantage may be bigger, because the overlap between the process technology in IDM and IFS is larger than between foundry and memory for Samsung. Second, Intel’s lead in advanced packaging is a powerful way to make up for some of the gap in process technology. Third, culture is central to any company’s success. If management can rebuild Intel’s former “Grovian” culture of measurements and results, business performance will greatly improve. Fourth, catching up on process technology. Keep a close eye on Intel bringing out five nodes in four years. They say they are currently on schedule. Finally, improving financial discipline. Zinsner’s last job was in semiconductor memory, which is a brutally competitive and cost-focused industry. If he can bring the financial focus and discipline from memory to Intel, they will save billions of dollars per year in spending. Overall, the arrogance that has concerned me from day 1 of Gelsigner is still there. It could just be the confidence and belief he has in Intel. The company’s strategy is the correct on, in my view. They are setting out to do all the things they need to do to turn Intel around and make it a leader again. Intel is the only company in the world that can catch up with TSMC and Samsung at the leading edge of foundry in the next five years. For Intel to be a good investment they don’t have to completely close the gap. They just have to be mostly successful for Intel to be a winning turnaround story.
-S. Hughes (no INTC position)