Intel sees AI boost in Q2
The boom in machine learning and AI in the data centre has boosted Intel’s results in Q2 despite continuing weakness in the enterprise and Chinese markets.
The company saw revenue of $12.9bn, down from $15.3bn on the same quarter last year and $19.6bn in 2021 but still at the top end of the estimates. This was $900m above the midpoint of the guidance but was partially offset by continued weak demand and higher inventory levels in the Network and Edge markets,
“In Q2, we began to see real benefits from our accelerating AI opportunity,” said Pat Gelsinger, Intel chief executive officer. “We believe we are in a unique position to drive the best possible TCO (total cost of ownership) for our customers at every node on the AI continuum. Our strategy is to democratize AI – scaling it and making it ubiquitous across the full continuum of workloads and usage models.”
“While we expect continued improvement to global macroeconomic conditions, the pace of recovery remains moderate. We will continue to focus on what we can control – prioritizing investments critical to our IDM2.0 transformation, prudently and aggressively managing expenses near-term, and driving fundamental improvements to our cost structure longer term,” he said.
A key change for Intel is the focus on open systems.
“We are championing an open ecosystem with a full suite of silicon and software IP to drive AI from cloud to enterprise, network, edge and client, across data prep, training and inference, in both discrete and integrated solutions,” he said.
“Intel Foundry Services, or IFS, positions us to further capitalize on the AI market opportunity as well as the growing need for a secure, diversified and resilient global supply chain. IFS is a significant accelerant to our IDM 2.0 strategy, and every day of geopolitical tension reinforces the correctness of our strategy.
“The building out of “Silicon Junction” in Magdeburg is an important part of our go-forward strategy, and with our investment in Poland and the Ireland sites, we already operate at scale in the region. We are encouraged to see the passage of the EU Chips bill supporting our building out an unrivalled capacity corridor in Europe,” he said.
“More importantly for our customers, it provides choice, leading-edge capacity outside of Asia and, at 18A and beyond, what we believe will deliver leadership performance. We are executing well on Intel 18A as a key foundry offering and continue to make substantial progress against our strategy.”
He points to Boeing and Northrop Grumman will joining the Rapid Assured Microelectronics Prototypes – Commercial (RAMP-C) programme along with IBM, Microsoft and Nvidia to ensure domestic access to next-generation semiconductors. “RAMP-C continues to build on recent customer and partner announcements by IFS, including MediaTek, Arm and a leading cloud, edge and data centre solutions provider,” he added.
However the company still has to deliver on its first EUV process technology with the 4nm Intel 4 process. This essentially complete with production ramping with the launch later this year of the chiplet-based Meteor Lake devices with a dedicated AI chiplet, Intel AI Boost, with an integrated neural VPU (vision processing unit).
“For the remaining three nodes I would highlight: Intel 3 met defect density and performance milestones for Q2, released PDK1.1, and is on track for overall yield and performance targets. We will launch Sierra Forest (SRF) in 1H’24 with Granite Rapids (GNR) following shortly thereafter, our lead vehicles for Intel 3. On Intel 20A, our first node using both RibbonFet and PowerVia, ARL, a volume client product, is currently running its first stepping in the fab. We will be the first to implement backside power delivery in silicon two-plus years ahead of the industry.
“In addition, back-side power improves ease of design, a major benefit not only for our own products, but even more so for our foundry customers. On Intel 18A, we continue to run internal and external test chips and remain on track to being manufacturing-ready in 2H of 2024.”
Like all the chip makers, Gelsinger sees recovery later this year. “We see a sustained recovery in the second half of the year as inventory has normalized. Importantly, we see the AI PC as a critical inflection point for the PC market over the coming years that will rival the importance of Centrino and Wi-Fi in the early 2000s, and we believe that Intel is very well positioned to capitalize on the emerging growth opportunity.
“We remain positive on the long-term outlook for PCs, as household density is stable to increasing across most regions and usage remains above pre-pandemic levels.”
This is despite licensing its NUC small PC business to Asus, one ofe the nine businesses that Intel has sold, generating $1.7bn.
In the datacentre, he sees the Gaudi chip as key. “Our pipeline of opportunities through 2024 is rapidly increasing and is now over $1 billion and continuing to expand with Gaudi driving the lion’s share. The value of our AI products is demonstrated by public instances of Gaudi at AWS and new commitments to our Gaudi product line from leading AI companies such as Hugging Face and Stability.ai, in addition to emerging AI leaders, including Indian Institute of Technology-Madras Pravartak and Genesis Cloud,” he said.
Quantum computing is also a key area, with the Tunnel Falls 12 qubit chip achieving 95% yield with voltage uniformity similar to chips manufactured under the more usual CMOS process, with a single 300 millimeter wafer providing 24,000 quantum dot test chips.
“We strongly believe our silicon approach is the only path to true cost-effective commercialization of quantum computing as silicon-based qubit approach is a million times smaller than alternative approaches,” he said.
Despite spinning Mobileye out last year, the company sees that as key, including the development of radar chips.
“Mobileye continued to generate strong profitability in Q2 and demonstrated impressive traction with their advanced product portfolio by announcing a SuperVision eyes-on, hands-off design win with Porsche and a mobility-as-a-service collaboration with Volkswagen Group that will soon begin testing in Austin, Texas,” said Gelsinger.
“We continue to drive technical and commercial engagement with them: co-developing leading FMCW (frequency modulated continuous wave) lidar products based on Intel’s silicon photonics technology and partnering to drive the software-defined automobile vision that integrates Mobileye’s ADAS (advanced driver-assistance systems) technology with Intel’s cockpit offerings.”
The cost cutting continues. “We remain on track to achieve our goal of reducing costs by $3 billion in 2023 and $8 billion to $10 billion exiting 2025,” he said. “Our new operating model establishes a separate P&L (profit and loss statement) for our manufacturing group, inclusive of IFS and TD (Technology Development Group), which enables us to facilitate and accelerate our efforts to de-risk our technology for external foundry customers and fundamentally change incentives to drive incremental efficiencies. We have already identified numerous gains in efficiency including factory loading, test and sort time reductions, packaging cost improvements, litho field utilization improvements, reductions in steppings, expedites and many more.”
That included the sale of a minority stake in Austrian litho specialist IMS. “IMS has created a significant market position with multi-beam mask writing tools that are critical to the semiconductor ecosystem for enabling EUV technology,” he said.
