IBM Power E1150
The Power E1150 is the midrange IBM Power 11 server, machine type 9043-MRU, and IBM lists it for 8, 12, 16 or 24 Spyre AI accelerator cards ... the same card ceiling as the smaller S1124. It is also the one Power 11 model an IBM i shop cannot move production onto, because IBM does not support IBM i on it.
What is it?
The Power E1150 is IBM's midrange Power 11 server. IBM calls it machine type 9043-MRU. It is a 4U box that holds two, three or four processor modules, and you can add modules later instead of buying a new machine. Think of the Power 11 line as three truck sizes. The S1122 and S1124 are the delivery vans, the E1180 is the semi, and the E1150 is the mid-size truck in between. It replaces the Power E1050 from the Power 10 line.
What IBM lists for the Power E1150
| Item | What is listed |
|---|---|
| Machine type model | 9043-MRU |
| Size and sockets | 4U, with two, three or four dual chip modules |
| General availability | 8 July 2025, announcement letter AD25-0034 |
| Processor features | EPEX 16 core, EPEY 24 core, EPEZ 30 core, each adding 2 spare cores per socket |
| Clock speeds | 3.40 to 4.20 GHz on EPEX, 3.05 to 4.15 GHz on EPEY, 2.8 to 3.95 GHz on EPEZ |
| Maximum cores and threads | 120 cores, 8 threads per core, 960 threads |
| Memory | Up to 16 TB of DDR5 across 64 DDIMM slots |
| Memory bandwidth | 512 GBps peak per socket |
| Expansion slots | 11 PCIe slots, using 176 Gen4 lanes, or 64 Gen5 lanes plus 64 Gen4 lanes |
| Internal storage | 10 NVMe PCIe Gen4 bays |
| Operating systems | AIX and Linux, with or without the Virtual I/O Server |
| IBM i | Not supported |
Every figure here is from IBM Redbooks publication SG24-8589, the Power11 E1150 introduction and technical overview, first edition October 2025, except the availability date and announcement letter, which come from IBM's product lifecycle record for 9043-MRU. A dual chip module, or DCM, puts two processor chips into one socket. A DDIMM is a differential DIMM, the buffered memory card Power 10 and Power 11 use in place of a plain memory stick. PCIe is short for Peripheral Component Interconnect Express, the slot standard adapters plug into. NVMe is the fast flash drive standard. GBps means gigabytes per second.
Can it run IBM i?
No. IBM says so twice in the same book. Section 1.6 of SG24-8589 says IBM i is not supported on the Power E1150 server, and chapter 2 repeats it as a flat restriction. The machine runs AIX and Linux partitions under PowerVM, with or without the Virtual I/O Server. This is worth knowing early, because the E1150 sits in IBM's Spyre card tables right beside models that do run IBM i, so on paper it looks like a choice. If your production work is IBM i, the Power 11 machines that can hold it are the S1122, the S1124 and the E1180.
IBM wrote that restriction as "at the time of writing" in October 2025. It describes the machine as it stands, not a promise about every future release. Check IBM's current documentation for 9043-MRU before you rule the machine in or out.
How many Spyre cards does it take?
IBM's Spyre Accelerator for Power publication lists 9043-MRU at 8, 12, 16 or 24 cards per system, the same four steps it lists for the smaller S1124. One partition can use up to 24 of them. Each card carries 32 accelerator cores, so a full 24 card system holds 768 accelerator cores. The cards never go in the server body. They only run in an ENZ0 PCIe Gen4 expansion drawer fitted with ENZA accelerator fanout modules.
AI accelerator cores at the top card step IBM lists for each Power 11 model
IBM's hardware requirements table lists 8, 12, 16 or 24 cards per system on 9043-MRU, 9824-42A and 9856-42H; 8 or 12 on 9824-22A, 9824-22B and 9856-22H; and 8 to 48 in groups of 8 or 12 on 9080-HEU. Any single partition tops out at 24 cards no matter which machine it runs on, so the E1150 can fill one partition completely. Each card holds 32 accelerator cores. A drawer carrying 12 cards needs Red Hat AI Inference Server 3.5 or later.
Why does the midrange size matter for AI?
Memory bandwidth, mostly. A model answers a question by reading an enormous pile of numbers very fast, so how quickly data reaches the chip matters as much as the chip itself. It is the difference between a wide doorway and a narrow one when the room has to empty in a hurry. IBM moved the Open Memory Interface on the E1150 to 32 Gbps with DDR5, up from 25.6 Gbps on the Power E1050, and raised the memory bus to 4000 Mbps per channel. Peak bandwidth per socket went from 409 GBps to 512 GBps. The 16 TB ceiling matters too, because a Spyre card reads the server's own DDR5 memory instead of carrying its own.
Peak memory bandwidth per processor socket
Both numbers come from Table 2-2 of IBM Redbooks publication SG24-8589, which sets the Power E1050 beside the Power E1150. The Open Memory Interface, or OMI, is the fast serial link between the processor and each memory card. IBM lists the E1150 memory bus at 4000 Mbps per OMI channel against 2666 or 3200 Mbps on the E1050.
What do you need before the cards work?
Four things IBM requires before Spyre cards run on an E1150
- A Red Hat Enterprise Linux partitionIBM lists Red Hat Enterprise Linux 9.6, 9.8 and 10.2. The cards are driven from Linux. On this machine that is the only option anyway, because there is no IBM i partition to drive them from.
- An ENZ0 drawer with ENZA fanout modulesIBM supports the cards only in the ENZ0 PCIe Gen4 expansion drawer, and only when that drawer carries ENZA accelerator fanout modules rather than the standard ENZF ones.
- Enough memory and cores in that partitionIBM's stated minimum for the partition running the cards is 512 GB of memory and 10 cores.
- A container runtimeIBM runs the Spyre software stack in containers using Podman and Quadlet.
These are IBM's listed hardware and software requirements for the Spyre Accelerator for Power, current to the September 2026 enablement stack release. A drawer holding 12 cards also needs Red Hat AI Inference Server 3.5 or later.
Related Terms
Spyre Accelerator
IBM's PCIe AI inference card. On IBM Power 11 it carries 32 cores, 128 GB of memory, and more than 300 TOPS inside a 75 watt envelope.
IBM Power E1180
The Power E1180 is the high-end IBM Power 11 server, machine type 9080-HEU, and the only Power model IBM lists for more than 24 Spyre AI accelerator cards ... up to 48 per system. An IBM i shop can keep production on IBM i 7.4 TR12 or later and run large AI inference in a Red Hat Enterprise Linux partition on the same machine.
IBM Power S1124
The Power S1124 is the largest scale-out IBM Power 11 server, machine type 9824-42A, and IBM lists it for 8, 12, 16 or 24 Spyre AI accelerator cards ... but only when both processor modules are installed. An IBM i shop can keep production on IBM i 7.4 TR12 or later and run model inference in a Red Hat Enterprise Linux partition on the same 4U machine.
ENZ0 Expansion Drawer
The PCIe4 expansion drawer IBM documents for supported Spyre Accelerator for Power configurations on Power 11 systems.
Differential DIMM (DDIMM)
A DDIMM is the memory card an IBM Power 11 or Power 10 server takes in place of a plain DDR memory stick. It carries its own buffer chip and plugs into an Open Memory Interface slot, which is how a two-socket Power S1124 reaches the 2456 GBps memory bandwidth ceiling IBM lists for feeding data into an AI model.
AI Workload Placement
The decision about which system, operating environment, and partition executes an AI model or agent, separately from the IBM i application supplying its data.