The Intel Corporation held a Second Quarter 2023 Earnings Conference Call, and John Pitzer, Corporate Vice President of Investor Relations, introduced the program. Pat Gelsinger, CEO, and David Zinsner, CFO, gave brief remarks, followed by a Q&A session. Pitzer noted that the earnings release and presentation were available on the investor website, and also warned that there were forward-looking statements and non-GAAP financial measures in the discussion. Gelsinger reported that the second quarter results exceeded expectations.

Intel is rebuilding customer confidence through effective execution across their process and product road maps. This strength in client and data center and cost savings has allowed Intel to return to profitability. Intel is focusing on AI to drive the best possible TCO for customers across the AI continuum and is driving an open ecosystem with silicon and software IP to do so. Intel Foundry Services is expanding their scale, accelerating their ramps and providing choice, leading edge capacity outside of Asia. They are making substantial progress against their 18A foundry strategy.

In the second quarter of 2021, Intel announced that Boeing and Northrop Grumman would join the RAMP-C program and made progress on two significant 18A opportunities. They are investing in manufacturing capacity and building out facilities in Germany, Poland, and Arizona. Intel's process and product road maps are on or ahead of schedule, and they are on track to regain transistor performance and power performance leadership by 2025.

Intel 7 and Meteor Lake are complete and Intel 4 is in production. Intel 3 is on track for its yield and performance targets and will launch Sierra Forest and Granite Rapids in 2024. Intel 20A is running its first stepping in the fab and will be the first to implement backside power delivery. Intel 18A is running test chips and will be manufacturing-ready in 2024. Intel's client business exceeded expectations and gained share in Q2 due to a modest recovery in consumer and education segments, as well as strength in premium segments. Intel is also partnering with Ericsson to utilize Intel's 18A process technology for its custom 5G SoC offerings.

Intel is well-positioned to capitalize on the emerging growth opportunity in the PC market with the AI PC, 13th Gen Intel processor family, and Meteor Lake. Meteor Lake will feature a dedicated AI engine, Intel AI Boost, and improved power, efficiency, and graphics performance. Intel is also ending direct investment in the NUC business but will continue to scale effectively through the recently announced ASUS partnership. In the data center, the 4th Gen Xeon Scalable processor is showing strong customer demand and is poised to ship its 1 millionth unit soon.

In Q2, Google Cloud announced the general availability of 4th Gen Cloud Instances and saw progress with 4th Gen AI acceleration capabilities. MLCommons published MLPerf training performance benchmark data showing 4th Gen Xeon and Habana Gaudi2 as strong open alternatives competing on both performance and price. Intel 3's Emerald Rapids 5th Gen Xeon Scalable is set to launch in Q4 of '23, Sierra Forest in the first half of '24, and Granite Rapids shortly thereafter. Clearwater Forest, the follow-on to Sierra Forest, will come to market in 2025. Intel is collaborating with Boston Consulting Group to deliver enterprise-grade, secure and responsible generative AI.

Intel performed better than expected in Q2, but there was still a soft consumption TAM for servers due to enterprise and rest of world weakness. They expect Q3 server CPUs to decline before recovering in Q4, but they are well positioned to gain share in 2024 and beyond due to the surging demand for AI products and services. Intel has a pipeline of opportunities worth over $1 billion, and they have shipped their test chip, Tunnel Falls, which was fabricated with a 95% yield rate with a single 300-millimeter wafer providing 24,000 quantum dot test chips.

Intel is focused on developing silicon-based quantum computing technology, and their Programmable Solutions Group, NEX, and Mobileye are all seeing strong demand in markets like industrial, auto, and infrastructure. In Q2, Intel announced the Intel Agilex 7 with the R-Tile chiplet, and they demonstrated the industry's first vRAN solution running on the 4th Gen Intel Xeon Scalable processor. Mobileye continued to be profitable and announced a design win with Porsche and a mobility-as-a-service collaboration with Volkswagen Group.

Intel has been driving technical and commercial engagement with Mobileye, co-developing FMCW and LiDAR products based on Intel silicon photonics technology, and aiming to drive the software-defined automobile vision. The company is also on track to reduce costs by $3 billion in 2023 and $8-$10 billion exiting 2025. Intel has identified numerous gains in efficiency and agreed to sell a minority stake in the IMS Nanofabrication business to Bain Capital, which has a long history of partnering with companies to drive growth and value creation. The IMS Nanofabrication business is critical to the semiconductor ecosystem for enabling EUV technology and will become even more critical with the adoption of high NA EUV in the second half of the decade.

Intel's second quarter saw strong business results, with revenue of $12.9 billion and a gross margin of 39.8%, beating expectations. Intel returned to profitability due to revenue strength, better gross margin, and disciplined OpEx management. Intel is continuing to focus on investments and cost structure improvements as part of its IDM 2.0 transformation.

Intel reported Q2 operating cash flow of $2.8 billion, up $4.6 billion sequentially. Net inventory was reduced, accounts receivable declined, and net CapEx was $5.5 billion. Intel's actions, such as the completed secondary offering of Mobileye shares and the upcoming investment in IMS Nanofabrication, will generate more than $2.4 billion of cash and $35 billion of shareholder value. CCG delivered revenue of $6.8 billion, up 18% sequentially, with operating profit of $1 billion, an improvement of more than $500 million sequentially. DCAI revenue was $4 billion, up 8% sequentially, with the Xeon business up double digits sequentially. Data center CPU TAM contracted in the first half of '23.

Intel experienced a decline in CPU TAM in Q2 due to a slower-than-anticipated recovery in China and across enterprise markets. DCAI had an operating loss of $161 million, but FPGA products delivered a third consecutive quarter of record revenue with another record quarterly operating margin. NEX revenue was $1.4 billion, down significantly in comparison to a record Q2 '22, and operating loss of $187 million improved sequentially. Mobileye had revenue of $454 million, roughly flat sequentially and year-over-year, with operating profit improving sequentially to $129 million. Intel Foundry Services revenue was $232 million, up 4x year-over-year and nearly doubling sequentially. Operating loss was $143 million with higher factory start-up costs offsetting stronger revenue.

Intel reported another strong quarter of disciplined spending with operating expenses down 14% year-over-year. They are on track to achieve $3 billion in spending reductions in '23, and have exited 9 lines of business, with a combined annual savings of more than $1.7 billion. For Q3, Intel expects revenue of $12.9-$13.9 billion, a gross margin of 43%, a tax rate of 13%, and EPS of $0.20. They are also investing in manufacturing capacity based on customer demand, and have closed agreements with governments in Poland and Germany which include capital incentives.

Intel expects net capital intensity in the mid-30s as a percentage of revenue in the next 18 months, with capital offsets in the range of 20-30%. Despite a slower-than-expected recovery in key markets, Intel is confident in its ability to capitalize on AI use cases and remain focused on execution and creating value for stakeholders. Ross Seymore was the first caller to ask a question about the data center side of Intel's business.

Intel's CEO Pat Gelsinger reported that the company had strong execution in the data center, with the Sapphire Rapids reaching 1 millionth unit in the next couple of days and the 5th Gen E-core, P-core with Sapphire and Granite Rapids performing well. However, there is some softness in the market due to big cloud customers prioritizing AI training environments, which is a near term surge that Intel expects to balance over time. Gen 4 Xeon has significant AI capabilities that will be enhanced in the future road map.

In the second quarter, Intel saw a significant increase in their gross margin due to the revenue beat, and they expect to see similar improvement in the third quarter due to the expected revenue growth. Additionally, Intel is expecting to see increased demand for their Gen 4 products driven by AI use cases, and they are seeing a lot of momentum with their Gaudi product line, with over $1 billion in pipeline.

The company is expecting a modest decline in underloading in the third quarter due to a period charge and the cost of inventory. Pre-PRQ reserves will still have an impact, but they will be lower than in the second quarter. Meteor Lake will not be a pre-PRQ reserve in the third quarter as it is expected to launch this quarter. The company is optimistic about the long-term outlook of gross margins, as they will eventually reach process parity and leadership, eliminating the headwind of start-up costs.

David Zinsner explains that data center pricing is up 17% year-on-year due to the introduction of Sapphire Rapids, and that platform costs will be affected by the increase in core count. He also notes that the internal foundry model will provide significant long-term gross margin improvement, as it will reduce underloading charges and start-up costs for 5 nodes in 4 years.

Pat Gelsinger explains that in Q3 they are expecting to see a decrease in the total addressable market due to inventory issues, weaker enterprise sales and the stronger spending on accelerators. Gelsinger notes that their overall position is strengthening due to the improved products and AI capabilities, as well as the increasing use of CPUs for Graph Neural Networks and Google's AlphaFold.

Pat Gelsinger believes that Artificial Intelligence (AI) will expand the server Total Addressable Market (TAM) due to productivity gains. He cites virtualization as an example of a technology that drove new workloads rather than destroying the CPU TAM. Gelsinger further explains that AI will be used in a variety of applications, ranging from 10 megawatts of power used for training models to 10 microwatts of power used in hearing aids. He believes that with Meteor Lake ushering in the AI PC generation, AI will be present in every application.

Intel is working to build AI into their products and believes in democratizing AI by making the software stack open. They view AI as one of the superpowers driving the trillion-dollar semiconductor opportunity. They are making good progress on their 5 nodes in 4 years, which culminates in 18A. PowerVia, the backside power, is a couple of years ahead of any other alternative in the industry and they have received a good response to it from customers, as evidenced by the Ericsson announcement.

David Zinsner discussed the two major foundry customers they had this quarter, their progress in producing 5 nodes in 4 years, and their success in advanced packaging technologies. He also noted that Intel managed their CapEx better than expected, resulting in more evenly distributed first half versus second half spending and improved free cash flow.

Pat Gelsinger speaks about the capital offsets they have been receiving from SCIP and how they will be getting more from CHIPS incentives, investment tax credits, and the EU Chips Act. Gelsinger also mentions their commitment to the $1 billion pipeline and merging the GPU and Gaudi road maps into Falcon Shores in 2025. He then allows Dave to add to the conversation.

Pat mentioned that the accelerator pipeline is now over $1 billion and growing rapidly, with Gaudi2 currently shipping volume product and Gaudi3 shipping volume product for next year. He also said that Falcon Shores will be released in 2025 and they are already working on Falcon Shores 2 for 2026. He mentioned that they are improving their software stack by adding FP8 and PyTorch 2 support, and expanding the language models and programmability. He also mentioned that they have Gaudi3 wafers and are expecting to give positive updates in the future. Finally, Pat mentioned that AI could have a Centrino moment next year with PCs.

Intel plans to bring AI capabilities to the client with its native AI capabilities in the second half of the year. This includes a neural engine and will be delivered in volume. AI applications such as real-time language translation, real-time transcription, automation, inferencing, relevance portraying, generated content, gaming environments, real-time creator environments, and new productivity tools will be enabled on the client, as they cannot be round-tripped to the cloud due to latency, bandwidth, and cost structures.

Intel expects to lead the way in democratizing AI at the client level and edge, which will drive the total addressable market. In the client business, Intel is confident in the outlook due to healthy inventory levels and solid demand signals from OEMs and channel partners, as well as gaining market share in Q2.

David Zinsner explains that the lack of strategic sales in Intel's 10-Q filing is an indication of how little the company did in terms of strategic purchases. In response to Srini Pajjuri's question about Intel's strategy for the AI custom silicon market, Pat Gelsinger explains that they have a good pipeline of foundry customers for 18A and are engaging with multiple opportunities to do their own versions of AI accelerators. Additionally, Intel has the strength of IDM 2.0 and advanced packaging to participate in these areas.

In the second quarter, Intel's team was credited for reducing inventory by $1 billion and reducing their days sales outstanding on accounts receivable to 24 days. This resulted in an improved free cash flow from Q1 to Q2. Intel still expects to reach breakeven free cash flow by the end of the year.

John Pitzer asked Aaron Rakers if he had a quick follow-up to his question about gross margin. David Zinsner answered that the PRQ and underload impacts were largely as expected, although the absolute dollars were less due to a higher revenue number. He also said that these impacts would be lower in the third quarter. Lastly, Aaron asked a question about the AI narrative, specifically Gaudi in the pipeline build-out.

Pat Gelsinger confirmed that Intel is expecting to see their Gaudi GPU deployed in hyperscale cloud providers and competing against larger competitors. They have already made Gaudi instances available on AWS, and have 1,000 customers taking advantage of their Intel Development Cloud. Intel had a good quarter, exceeding expectations on top and bottom line, and they are looking forward to continuing to accelerate their business and seeing margin improvement in the second half of the year.

The speaker thanked their team for a successful quarter and invited investors to an Innovation Q&A track in September. They highlighted the operational improvements, cost savings, and execution progress that were made. The conference was then concluded.

This summary was generated with AI and may contain some inaccuracies.