Intel posted what CEO Bob Swan characterized as the "best quarter in company history" today, exceeding its revenue targets by $1.2 billion on its way to a record $19.2 billion in overall revenue. The impressive results were fed by a record $9.5 billion in revenue for Intel's data center group, which now comprises 49% of the company's revenue (another record).
Meanwhile, client computing (aka desktop PCs) revenue fell five percent on the year, which Intel attributed to a 'tough compare' to last year's strong performance. However, Intel's volume in desktop processors dropped 10% year-over-year, but higher average selling prices (ASPs) from high-performance products helped buoy revenue.
AMD's surging presence in the do-it-yourself market may be partially responsible for the decline, but we'll have to wait until AMD's earnings report next week (and concrete market share data) to assess the impact.
Intel noted that one of the key challenges for the company's margins next year would be the "competitive impact" on average selling prices, alluding to price pressure from competitors. Swan elaborated that "we've tried the best we can take into account competitive dynamics as we exit this year and going into next year in our quest to play a bigger role in our customers success. We're going to compete to protect our position and expand the role we play." It's easy to assume the statements refer to pricing pressure due to AMD's expanding portion of the desktop PC market.
Swan announced during the call that Intel had powered on the company's DG1 GPU, its first discrete graphics card based on the 10nm process and Xe architecture, for the first time during the quarter, which "is an important milestone." The 10nm DG1 is expected to launch next year, and Swan also revealed that the company is on track to deliver its 7nm discrete GPU in the fourth quarter of 2021. That indicates that Raja Koduri's graphics group is on schedule with a regular cadence of products in the pipeline.
Intel has increased 14nm wafer starts by 25% this year, and plans to increase both 14nm and 10nm production by another 25% next year. In spite of these capacity increases, Swan said the company is still struggling to meet demand. Swan said the shortage of 14nm production capacity would extend into the fourth quarter of 2019, noting that "we're letting our customers down and they're expecting more from us."
Intel's OEM partners are now shipping eighteen 10nm Ice Lake designs into the market, out of 30 planned models. The company is producing 10nm chips in its Israel and Oregon fabs, with additional capacity coming online in Arizona soon. Intel plans to return to a two-year to 2.5-year node shrinking cadence with 7nm and following nodes and is "well down the engineering path on 5nm" development.
Swan also seemed to concede, for the first time, that the company no longer holds the lead in process technology, remarking that "We're working to recapture process leadership."
Intel's data center group remains a bright spot. Intel said that it had sold 23 million Cascade Lake data center processors to date and that a high mix of premium chips led to a 9% increase in ASPs. Again, this richer product mix offset a 6% decline in platform volumes. Swan also mentioned that two customers had selected the company's rare 9200-series Cascade Lake-AP Xeon processors for volume deployments.
Other notable gains include Intel claiming it has now taken the lead in the networking market with $5 billion in revenue forecast for this year, and that its IoT division has now broken $1 billion in quarterly revenue for the first time. AI-centric revenue also topped $3.5 billion, which is an important achievement for Intel as it continues its full-court press on multiple segments of the AI market.
Intel's gross margin also fell to 58.9%, down appreciably from 64.5% during Q3 2018. The company chalked that up to ongoing investments in its transition to the 10nm node, among other factors.