Intel returns to long play, not bad play

The supply chain holds back server activity, not just the way you think of it. Yes, there is a limited supply of manufacturing and packaging capacity for server-class processors based on the most advanced semiconductor nodes. It’s in less obvious areas, such as the availability of power supplies and various types of controllers and peripherals, that OEMs and ODMs need when building a server.

“Believe me, we would ship a lot more units if we weren’t limited by the supply chain of these other industry components,” said Pat Gelsinger, the CEO who joined Intel in January for the save from itself. and let him return to what he once was. “Our customers, both cloud customers and OEMs, have very large backlogs that they are pushing us to satisfy aggressively. But we’re really limited by these “match sets” as we call them in the industry. “

Another constraining factor for the Intel data center group, which still exists in the company’s finances despite a reorganization in June, has been the reduction in shipments of server chips and other data center components. data to China. Gelsinger said there had been recent regulatory changes in China – what he called a “new regulatory environment” which actually affected shipments of the SP Xeon “Ice Lake” in the Middle Kingdom. for supercomputers and cloud builders, and, we believe, will also reduce those of future Xeon SP “Sapphire Rapids” scheduled for production from Q1 2022 and volumes and announcement in Q2 2022, according to Gelsinger. Who added that Intel had “a particularly high market share in the Chinese cloud market”.

We suspect that is changing, and for many reasons including the forays of AMD and Ampere Computing and the desire of hyperscalers and cloud builders to control their own destiny, like the Arm server chip. Custom Yitlan 710 from Alibaba, which we talked about. pretty much ahead of its launch last week and we’ll dig deeper as long as we can get the streams and speeds. And so we suspect that Intel has more than a regulatory change in China, including the ever-expanding list of entities where certain types of chips are blocked for sale by the Industry and Safety Bureau of the. US Department of Commerce.

In the quarter ended in September, which disappointed Wall Street more for its forecast than the actual figures for the quarter, but both weaker than expected, Intel raked in $ 19.2 billion in revenue, up 4 , 7%, and thanks to a gain of 1.71 billion dollars. on equity investments, was able to post a net profit of 6.82 billion dollars, up 59.7%.

In the Data Center group, Gelsinger was happy to report that between April and September, the company shipped over one million SP Ice Lake Xeons, and the ramp was progressing so that another million units would be shipped in the fourth quarter, which is a factor of 2X more units in a three month period. What we found interesting, however, was that the mix for Ice Lake Xeon SP sales was more for the HCC variants – those with 8 to 28 cores – rather than the XCC variants – which range from 16 to 40 hearts. (There is apparently no LCC variant with the Ice Lake chips, as there were with most previous generations of Xeon and Xeon SP processors.) Another mix change that brought down selling prices means of chips in Data Center Group (and therefore revenue) was the shift from processors to SoCs, according to George Davis, chief financial officer of Intel, who announced he was retiring during the call with analysts at Wall Street.

Data Center Group sales were just under $ 6.5 billion, up 10%, and operating profit rose slightly less than 8.1% to $ 2.06 billion. The operating profit was 31.7% of sales, which is much better than the 22.9% it was in the first quarter of 2021 and a little better than the 30.1% of the second. quarter of 2021, but we are far from nearly 50%. operating profit that Data Center Group had while there was no competition in the market in the glory days of late 2009 to early 2020.

Hyperscalers and cloud builders – what Intel calls cloud service providers – spent 20% less on Intel chips and other components in the third quarter, and with them accounting for such a large chunk of Intel’s sales, when they slow down their spending, it really has a dramatic impact on Intel’s overall sales. And so, even though corporate and government customer spending increased by 70% – down from a low number a year ago, mind you – and even though communication service providers (telcos and others) increased by 18% , the entire Data Center group sales were more subdued than they could have been otherwise. Intel didn’t say the impact came from regulatory changes in China, but we think it crossed the fine line between “important” and “hardware” – but that’s just a hunch.

As usual, IoT + Mobileye Group and Programmable Solutions Group (formerly known as Altera) that sell FPGAs couldn’t close the gap, but IoT + Mobileye Group at least contributed significantly to revenue growth and really helped operating margins, as you can see:

We have been tracking Altera / PSG activity separately for some time now and frankly we expected this activity to grow much faster than it has. In the third quarter ended in September, PSG achieved revenue of $ 478 million, up 16.3%, which is more than three times the overall growth of Intel, and the result of operations increased 90% to $ 76 million. If you spread the net profit proportionally to revenue and adjust the overhead, the PSG could be something like $ 100 million in net profit, which would be 20.9% of revenue, which is the profit ratio. net to the revenues Altera had in 2013 and 2014, just as Intel started looking to acquire Altera and it looked like FPGAs would be more popular than they have been.

Each quarter, we try to determine what Intel’s ‘real’ data center activity looks like, including the parts of its various groups that actually sell products and services in the data center with the reported revenue. of the data center group. Here’s what our model looks like for all of 2020 and the first three quarters of 2021:

And for those of you who like your data to be presented visually to better see trends, here’s a plot that goes back to 2015, when we were able to build this model for the first time:

After our allocations, which we admit are based on educated guesses, we estimate that revenue from Intel’s “real” data center business actually increased by 11.1%, to 8%. $ 37 billion, and operating profits rose 31.2% to $ 2.67 billion, which is working. at 31.2% of sales.

Looking ahead, senior Intel executives say they expect revenue of $ 74 billion in 2022, with a compound annual growth rate of 10% to 12% over the past four to four. next five years, and gross margins reaching 51% to 53%. in the next two or three years. Capital spending is pegged at $ 25-28 billion in 2022, which is roughly a year and a half of modern 7-nanometer factories, give or take, so don’t be impressed, be horrified to see it’s whatever money makes you.

Maybe we’re just getting cynical, but we think Intel is going to have a much tougher sledge than that. But Gelsinger isn’t just a technologist and a manager, he’s an optimist. And he was happy to tell Wall Street that the next five semiconductor manufacturing nodes – 10 nanometers SuperFIN have already come out, followed by Intel 7 (what we used to call 7 nanometers), Intel 4 (what we used to call. 5 nanometers), Intel 3, Intel 20A, and Intel 18A – all of which are slated for release in the next four years, are on time or ahead of schedule.

“Relatively speaking, we are closing the gap with the industry, probably even faster than I would have imagined just a quarter ago,” Gelsinger said. “And as a result, these investments will produce higher quality products with higher prices and margins faster than we would have expected even a quarter ago.”

Taiwan Semiconductor Manufacturing Co is reportedly having issues with its 3-nanometer process, so maybe Intel’s turn to be in the driver’s seat for a while. Stranger things have happened. Like Intel almost forgetting that it had to be at the forefront of chip manufacturing and then also design excellent chips. The former has always been more important than the latter when it comes to Intel processors, and Intel may have remembered in time not to lose its server business to the competition that filled the void it left. . Still, look at all these profits from 2009 to 2020.. . . It is an object lesson about short-term gain versus long-term gain. Now, with Gelsinger at the helm, Intel is back to the long game. At least it’s not the bad game.

Source link

About Mariel Baker

Check Also

Will Nepo’s supercomputer give him an edge over Carlsen for the world title? | Chess

IIt would count as one of the most seismic shocks in modern chess history if …