IBM: Is The Cloud Swallowing the Mainframe? Asks Bernstein
Bernstein analyst Toni Sacconaghi today continues his assault on shares of IBM (IBM ), posing the question of whether the company’s mainframe business can bolster its profit, and concluding that No, in fact, mainframes may be a source of weakness. Computing tasks that might have used one are increasing going to cloud computing servers.
A key near-term debate among investors is whether the anticipated release of a new mainframe can help IBM achieve a back-end-loaded second half.
The immediate answer, he writes, is a refresh of the mainframe may add up to 19 cents in earnings per share in the second half of this year, which is “not enough to get us comfortable with IBM’s fiscal 2017 EPS guidance.
Mainframes are just 3% of total revenue, and 2% of profit, he notes, annually, but the whole “platform” of a mainframe, including storage, software, support contracts, and services that go with it, were nearly a quarter of IBM’s revenue last year, and 40% of profits.
Moreover, that mainframe cluster of offerings has an annuity quality that makes it attractive:
Importantly, mainframe platform revenues and profits are largely (
80% to 90%) driven by the installed base of mainframes and therefore provide a strong base of recurring/annuity-like business for the company. Specifically, we note that mainframe software (both operating systems and middleware – which we estimate amounts to nearly $10B in annual revenues) is sold on a monthly subscription basis, meaning that it is highly stable and tied to IBM’s 20 year+ installed base of mainframes. Similarly, mainframe’s lucrative maintenance/support and financing businesses are also largely installed base driven.
He figures if IBM delivers a new model by “mid-august,” it can add 16 cents of EPS for the rest of the year, along with sales of “high-end storage” to go with it.
But that would mean IBM’s fiscal year profit of “at least $13.80” would still be “back-end loaded.” Instead of having to get 63% of that total from the second half of the year, it would have to get 61.6%, still a tall order that makes him uncomfortable.
Worse, we appear to be seeing the start of a negative inflection point” for mainframes, says Sacconaghi.
He opines it’s the cloud:
The notable decline in mainframe peak and trough revenues points to incremental structural pressure on the mainframe. While it is possible that we are seeing an accelerated rate of migrations of existing apps off of the mainframe platform, we believe that the more likely cause is that new workloads or incremental volumes from existing workloads that would have historically gone to a mainframe are now either moving to the cloud, or potentially to a lower cost architecture. We think this might particularly be the case with Linux workloads – when first introduced in 2000, the ability to place Linux workloads on mainframes provided companies with a way to consolidate workloads from poorly utilized x86 servers (one application per server and utilization likely 20% or less) to an existing mainframe with native virtualization capabilities. However, today enterprises have several choices for consolidating workloads, most notably the public cloud or using VMware (or other software) to virtualize x86 servers and significantly boost utilization rates (and cut down on server sprawl).
If mainframe sales “decline steadily” in coming years,” it could hurt profit: if mainframe is
40% of company profits, and mainframe hardware falls in half over the next 15 years (about a 4% decline per year), this would negatively impacting IBM’s installed base of mainframes by about 25%, and impact mainframe profits by potentially 30%.
If mainframes are 40% of company profits, then company profits could be impacted by 12% over the timeframe or nearly 100 bps per year.
IBM shares today are down 29 cents at $154.11.