Power9 Enters The Long Tail
January 27, 2020 Timothy Prickett Morgan
There is no way to sugar coat this, so I will just say it: It is going to be a long year for the Power Systems platform this year, and I will be so pleasantly surprised and happy to be wrong about that. But I don’t think I will be.
The Power Systems platform, as we have explained in the past, had a great 2018, with revenues up 9.5 percent to just a tad over $2 billion as far as we can tell from our financial model, which we have painstakingly developed. Now, mind you, the Power Systems market exited the Great Recession in 2009 with $5.11 billion in sales, and went into a gradually increasing decline, with sharp declines in 2013 and 2016, to flatten out at $1.84 billion in both 2016 and 2017 to see that bounce in 2018. IBM had seven straight quarters of revenue growth starting in Q4 2017, with 6 percent growth in Q1 2019 and 1 percent in Q2 2019, but sales dropped by 27 percent in Q3 2019 and 24 percent in Q4 2019.
To be fair, as Jim Kavanaugh, IBM’s chief financial officer, explained to Wall Street in a conference call last week to go over the financials for the fourth quarter, IBM was booking revenues for high-end Power9 iron in the third and fourth quarters of 2018, so the compares were brutal. There are only so many big IBM and AIX shops in the world – thousands but not tens of thousands – and they behave more like IBM System z shops in their buying patterns than anything else.
IBM does not provide financial figures for Power Systems sales, but said that revenues were down 24 percent as reported and down 23 percent at constant currency. When we do the math on that, we think IBM booked $457 million for servers and around $80 million in revenues for Power Systems machinery that was used in IBM’s DS-series storage appliances that should be counted as Power Systems revenues. (But is not in the accounting methodology that IBM uses with its reports to Wall Street and the US Securities and Exchange Commission.) Add that up, and Power Systems sales were off 20.9 percent to $537 million in Q4 2019.
Here is what the quarterly results look like over time according to our model:
Here is what happens if you collapse it down to annual sales instead of quarterly sales:
It is hard to figure out what will happen in the Power Systems business looking ahead, but as you can see from the annual chart, the launch of the Power7 processors in 2010 and the Power8 processors in 2014 did not stop the radical change and move away from RISC architectures running Unix or proprietary platforms in the past decade. Linux on Power has helped mitigate some revenue declines, but we are nowhere near getting back to the $5 billion baseline of revenues in Power Systems that I think of as historically safe. But Moore’s Law is responsible for most of this revenue decline, and in 2019 and 2020, the long tail of Power9 shipments, which will not go to zero but will slow down significantly as the market awaits the Power10 chip and its anticipated large price/performance and performance increases, will be trying.
How low it will go, who can say? What is important is that IBM is selling more System z capacity than it has ever done in a quarter with the launch of the System z15 machines, which had their first full quarter of shipments in the fourth quarter, and revenues were up 63 percent and MIPS capacity was very high (we have no idea what the growth rate was because IBM doesn’t talk about it anymore). Those mainframe sales have made Big Blue’s overall Systems Group sales very healthy, and its operating incomes, too, and this is what matters. Each IBM platform in its turn helps fuel the business, and IBM can continue to invest in both platforms. Ditto for adding Red Hat Enterprise Linux and its OpenStack and OpenShift extensions, which are not reported in the Systems Group numbers.
In the fourth quarter, IBM’s overall sales were down a fraction of a percent to $21.78 million, but gross profits nearly doubled to $3.67 billion. IBM had a much lower tax burden this time around, which accounted for most of the swing to the deeper black ink. IBM’s overall Systems group revenues were $3.24 billion, up 13.3 percent. Systems hardware revenues (including servers and storage) rose by 17.4 percent to $2.55 billion. Operating systems sales on IBM’s platforms grew by 9.5 percent to $490 million; this does not include Red Hat Enterprise Linux, which it obviously should but IBM wants to pad its cloud revenues elsewhere. The gross profits for Systems Group rose by 28 percent to $1.71 billion, and pretax income rose by 45.6 percent to $802 million. This is the most profitable quarter, relative to revenues, that IBM has turned in since Q4 2017.
IBM’s real systems business, as best as we can figure, is much larger that it lets on in its financial presentations. We try to suss that out every quarter to provide some comfort to those buying IBM’s systems so they don’t just bolt. Our model, which takes into account hardware, systems software, services, and financing for systems, rose by 15.9 percent to $8.85 billion, accounting for a very decent 40.6 percent of IBM’s overall revenues in Q4 2019. Operating profits, however, seemed to be under a little pressure despite the big bump in System z sales, and fell by 14.6 percent to $3.84 billion. For the full 2019 year, we reckon IBM’s “real” systems sales were $27.45 billion, up 5.7 percent, but operating profits were down 22.8 percent to $11.3 billion. The earlier three quarters of 2019 were not kind to IBM’s bottom line, but Big Blue plays the very long game here.
Which is a very good thing, indeed.