Updates - INTC - Readjusting Expectations
Summary
2Q22 results were very disappointing, heavily impacted by macro uncertainties.
We recap some recent history to help reflect on Intel's current status.
We review the outlook for Intel's different business groups in the context of the global macro economy.
We're still bullish over the long-term, though we've had to readjust expectations.
Big Picture Reflections
INTC's 2Q22 earnings result is a massive downfall for the firm. Similar to NVDA, INTC is signaling the eventual recession scenario that will affect consumer and enterprise chip demand. To understand where we are today, let's have a quick recap of the economic situation leading to today.
Mobile Overtaking PC
In the period leading up to 2011, global PC demand rose and peaked at 365 million. Coincidentally, it was also the year when INTC's Tick-Tock strategy, created by Pat Gelsinger, started to fall apart and chip giant's relentless effort to deliver Moore's Law began to falter. After 2011, INTC basically kept prices high with miniscule performance gains each year. At the same time, mobile phones were on the rise. Mobile SoC fabless vendors like QCOM, MediaTek, and AAPL quickly overtook INTC's prestige semiconductor leadership thanks to skyrocketing mobile device shipments.
Taking tablets and mobile phones together, in 2016, total shipments stabilised around 1.8 billion per annum, while PC shipments including desktop and laptop stabilised around 270 million. As virtually all mobile devices are Arm-based while PCs are x86-based, Arm chip volumes now overwhelm x86 by at least a factor of 6x.
Furthermore, if we factor in other Arm-based devices like IoT, auto, drones, and many other embedded devices, Arm's total shipment in 2021 was c. 30billion, while INTC's shipment was at around 300million at best (PC+server). This is a 100x volume difference.
Alongside R&D and manufacturing capabilities, scale is the most important factor in the success of semiconductor vendors. They need enough scale to cover the colossal front-end costs associated with cutting-edge nodes, such as the 3nm that cost $20bn+ to develop. At the current volume, it has transpired that INTC wasn't able to keep up with the pace of TSM, who manufactured the majority of Arm and non-x86 chips worldwide.
Luckily for INTC, the slow down in PCs had not turned its revenue growth rate negative, as it found growth in enterprise data centres. With the rise of internet companies and public clouds, the demand for server chips ballooned during the 2010s. With near-monopolistic leadership in high-end CPUs, INTC has grabbed 90%+ of this lucrative market as enterprise customers were willing to pay a 2x+ premium for more durable CPUs running 24/7 in production environments.
During this period, INTC pretty much abandoned its consumer PC R&D efforts. The performance gain each year was negligible, which in effect lengthened the refresh cycle because there was little point in consumers buying a new PC in any given year. This further suppressed consumer demand, and according to INTC's management, they now have over 600 million PCs older than five years that are still being used by consumers.
AMD's Comeback
This finally comes to an end in 2019 when AMD released the ZEN series CPU that reached parity with INTC. With lower cost and higher performance from AMD, INTC was forced to increase the core counts without increasing the price. As we are back to competition and innovation, the refreshment demand would pick up. For example, INTC's low voltage mobile 7th Gen i7 CPU had only 2 cores, while its 8th Gen i7 CPU had 4 cores in 2018. For standard voltage mobile i7 CPU, the core count increased from 4 during the 2010 to 2017 period, and then to 8 in 2020.
The refreshment cycle driven by resumed product performance/cost gains is further accelerated by COVID-19. During the outbreak of the pandemic, the demand for remote work and study reached a culmination, which revived demand for PCs/laptops.
This led to two consecutive increases in PC shipments in 2020 and then 2021. At its peak, many analysts predicted that 2022 would be another good year and that PC shipments may return to 2011 levels.
The second wave of PC demand supported much of INTC's earnings and offset some of the tough fight against AMD. AMD has the more efficient core architecture and leveraged TSM's fab to bring 7nm on time to market while INTC was stuck at 14nm, hence in a normalised environment AMD would have clearly outcompeted INTC. However, as the adage goes - a rising tide lifts all boats - the pandemic-induced demand gave INTC a lifeline.
Thus INTC accepted a lower gross margin (GM) and stayed competitive in PC. But for DC, AMD's chiplet is a game changer. It delivered unchallenged performance/cost relative to INTC to such an extent that price bundling and discounts could only slow but not stop AMD's traction. The worst moment came in 2021 when AMD launched its ZEN3 Milan series CPU for enterprise data centre servers. For the period between 2017 and 2021, AMD had been seeding the ground for its DC market. This is a slower moving market relative to consumer PC, as enterprises' top priority is stability rather than outright performance/cost. During this period, AMD successfully managed to run through stringent product testing and mature its chiplet technology for DC use cases.
Meanwhile, INTC tried to defend its market share by lowering the price with its IDM advantage. However, due to its less efficient architecture, it has to increase the size of the core to maintain a performance parity with AMD's, and this puts pressure on its GM even though as an IDM it doesn't have to pay for fab costs like AMD does.
Now Huge Macro Headwind for Chip Demand
As we are entering 2022, the macro economy started to deteriorate. Inflation has reached a critical state as the base effect is expected to normalise the YoY comparisons, thus putting the 'transitory' narrative under even more scrutiny.
The supply chain crisis, heightened demand for physical goods, and the subsequent Russian invasion of Ukraine introduced major shocks to the system. Inflation rising into March and beyond surprised economists who predicted a more moderate YoY CPI number. As a result, the Fed was forced into aggressive rate hikes to halt inflation by cooling demand.
This introduced a perfect storm to the economy. The previous few quarters of GDP growth was primarily driven by inventory build up in expectation of prolonged supply chain issues and heightened demand. Now multiple industries are suddenly overstocked with balance sheets rapidly deteriorating, as exemplified by Walmart and Target's latest ERs.
The earliest sign of weakness in the tech industry started with ads numbers related names like SNAP, META, and TWTR, which reported very weak demand from advertisers. Analysts then speculated that consumer spending is going to plummet and began cutting expectations for consumer names. Meanwhile, PC vendors are also in crisis mode again as they are overstocked. Thus, they cancel more orders for INTC in a U-turn to the originally planned order increases.
This results in weaker PC shipments, and lower margin as INTC needs to give more discounts to maintain sales, all amid higher costs as inflation rages on. INTC's bread-and-butter PC business thus delivered poor quarterly results and dragged overall results lower.
To rub salt into INTC's wounds, AMD's Milan SKUs began to ship massive volumes as it was adopted by DCs worldwide. At the same time, INTC's competitive SPR (Sapphire Rapids) encountered unexpected security issues during testing and INTC was forced to do further rounds of design, verification, manufacturing, and testing, which delayed the mass adoption by a further two quarters. As a result, INTC's second pillar, the DCAI (Data Centre & AI) group, is in bad shape as well.
The AXG (Accelerated Computing & Graphics) group is no better. INTC did launch its first alchemist GPU in China for laptops and then PCI-e cards for desktops. The initial feedback is that the performance is good, matching parity with NVDA's SKUs. However, in many games, INTC's GPU doesn't perform smoothly. In less popular games and older games, INTC's GPU faced a 50%+ performance decline due to driver compatibility issues. As a result, many potential buyers stayed on the sideline waiting for the software stack to mature. This resulted in higher inventory costs for INTC and weaker sales.
IFS sign more promising deals in 1H22, but similar to SPR and Granite Rapids, they will only meaningfully contribute to revenue in late 2024 and beyond. For signed customers, it includes MediaTek on Intel 16, QCOM on Intel 20A, AWS for advanced packaging, and even NVDA, who is now becoming more and more of a competitor to INTC.
IFS is super-capital intensive as INTC needs to build more capacities, not only for its own demand, but also for third-party vendors. INTC has to build numerous fabs to scale the shipping volume closer to TSM in order to compete effectively. This means that it needs to spend close to $100bn in capex.
With more product revenue to come in the future, INTC has to source more capital via partnerships. Recently, it announced a $20bn investment from Brookfield, a renowned infrastructure PE investor, to buy 49% stake in its new Arizona 20A fab. Brookfield is one of the smartest PEs doing infrastructure related investments. Therefore, one can sense that the smart players are also giving a nod to INTC's IFS future. Furthermore, Pat Gelsinger with his political savviness, is actively pushing for government deals. The new CHIPS act could save INTC's new fabs capex by 50% at least, and it is expected to be signed by Biden very soon. Meanwhile, various European countries, Germany and Italy in particular, are also eager to sign deals with INTC leading to up to 50% capex reductions via tax benefits and fiscal supports.