Gloomy second-quarter report: Intel is on the verge of being forgotten

Author: Zhou Yuan / Wall Street News

"If this continues, Intel will soon be forgotten." A research director at a US dollar fund told Wall Street News.

On August 1st, California time, Intel released its financial report for the second quarter of the 2024 fiscal year, showing that Intel's revenue in the second quarter was $12.833 billion (lower than the market's expected $12.94 billion), a 1% year-on-year decrease; a net loss of $1.654 billion, compared to a net profit of $1.473 billion in the same period in 2023, turning into a loss year-on-year; gross margin was 35.4%, lower than the 35.8% in the same period in 2023, and lower than the 41% gross margin in the first quarter of 2024.

On the same day, Intel's stock price fell by 5.5% to $29.05 per share; after the US stock market closed, Intel's stock price fell by 18.90%.

Intel CEO Pat Gelsinger analyzed, "The performance in the second quarter was below expectations, partly due to the company's decision to accelerate the production of Ultra AI PC CPUs. The investment in defining and driving the AI PC category will bring significant pressure on profits in the short term, but we believe it is worth it. The market share of AI PCs will grow from less than 10% currently to over 50% by 2026. The current investment will bring significant growth for us in the coming years."

According to the new business structure adopted in the first quarter of 2024, Intel's business is mainly divided into three major segments: Client Computing Group (CCG), Data Center and AI Business Group (DCAI), Network and Edge Business Group (NEX). These three parts are collectively referred to as "Intel Products"; Intel Foundry business is listed separately; Altera (formerly Intel's Programmable Solutions Business), Mobileye, and others are referred to as "All Other".

In the second quarter, the revenue of Intel Products in the three business segments was $7.41 billion (a 9% year-on-year increase), $3.05 billion (a 3% year-on-year decrease), and $1.3 billion (a 1% year-on-year decrease); the revenue proportions were 57.74%, 23.77%, and 10.13% respectively.

Among them, the CCG business achieved growth mainly because since December 2023, AI PC shipments have exceeded 15 million units, with shipments expected to exceed 40 million units by the end of 2024.

Intel Foundry revenue reached $4.32 billion, a 4% year-on-year increase, accounting for 33.66% of total revenue; revenue from All Other businesses was $0.968 billion, a 32% year-on-year decrease, accounting for 7.54%; among them, Altera achieved revenue of $0.361 billion, a significant 57% year-on-year decrease, and Mobileye revenue was $0.44 billion, a 3% year-on-year decrease.

Gelsinger also expressed disappointment: "Although we have achieved key product and process technology milestones, our financial performance in the second quarter was disappointing. Our revenue did not grow as expected, and we have not fully benefited from strong trends such as artificial intelligence. Our costs are too high, and our profits are too low." Of course, as the top leader, Kissinger has not forgotten to boost morale. He said, "(Although) the market trend in the second half of the year will be more challenging than expected, we are taking decisive actions to improve operational and capital efficiency with a new operating model, and accelerate the transformation to IDM 2.0."

Kissinger believes that through the above measures, coupled with the Intel 18A process technology to be launched in 2025, will help Intel regain its leading position in process technology, consolidate market advantages, improve profitability, and create shareholder value.

However, the current reality may still require Intel to continue to endure agony.

Along with the release of the second quarter financial report for 2024, Intel also announced a $10 billion cost-cutting plan (to be completed by 2025) and initiated a "slimming down" process of global workforce reduction by 15% (to be completed by the end of 2024), marking the largest layoff in Intel's 56-year history. As of June 29, Intel (excluding some subsidiaries) had a total of 116,500 employees. Based on a 15% layoff rate, Intel will lay off at least 17,000 employees this time.

More importantly, due to the poor market performance in the second quarter, Intel will suspend dividends starting from the fourth quarter of 2024 (until cash flow improves to a sustainable higher level), marking the first time since 1992.

In response, Kissinger explained, "We must align our cost structure with the new operating model, we need to fundamentally change the way we operate."

The performance outlook for the third quarter of this year is also disappointing for Intel. Intel expects revenue of $12.5 billion to $13.5 billion in the third quarter of 2024, far below analysts' expectations of $14.38 billion.

Intel's main product is CPUs. The company just announced at the end of July that it will launch its strongest AI accelerator card, Gaudi 3 (TSMC 5nm process), in the fourth quarter of this year. This is Intel's only AI server accelerator card, as Intel has not introduced any other AI accelerator chips.

Gaudi 3 will serve as Intel's flagship server AI accelerator card, competing with NVIDIA's flagship Hx00 Hopper architecture AI accelerator card.

However, because Gaudi 3 is scheduled to be launched in the fourth quarter, the predicted revenue scale is only $500 million, not comparable to NVIDIA, and far behind the $4.5 billion revenue scale formed by AMD's MI300 series AI accelerator cards (for the full year of 2024).

Technically, Gaudi 3 strangely only integrates HBM2e, instead of the more advanced and current mainstream HBM3 or HBM3e. Therefore, the maximum capacity stack available for Gaudi 3 is 16GB, with a total of 128GB of memory capacity across 8 layers. This is generally a "last resort" choice when there are no other options, but Intel clearly did not face any restrictions, so why only use HBM2e?

This indicates that Intel's internal organizational mechanisms are too slow to react, adjustments are too slow, and they have failed to seize the market opportunity after the release of ChatGPT-3.5 (released around November 30, 2022, some say January 23, 2023), and the technology has not shifted quickly. Even in 2024, it is difficult to see Intel regaining the technical and product advantages in the AI field as it did in the CPU industry in the past For Intel, the time left for them is indeed running out