HN Explorer

Ask HN: Why is IBM still around?

3 points by temp_acct_0xabc ↗ HN
I don't get it. They haven't put anything useful or successful out of the door in decades and they are still around. How do they do it?

I mean, more successful products and even market leaders such as AWS are feeling the pressure of the downturn (e.g. even AWS ached with the loss of many startups that downsized or went away due to recent SVB crisis). And yet here you have a leader-to-nothing dinosaur that despite all the crisis in the world is still alive. I honestly don't get it.

12 comments

[ 2.4 ms ] story [ 40.8 ms ] thread
jstx1
They offer products and services, companies pay for their products and services. When everyone else was inflating massively, they didn't, so now that the markets have turned they don't have a lot of deflating to do.
raverbashing
There's a lot of CxO type level people who still buy their marketing (which is not only about fancy ads)
brucethemoose2
They technically have some useful IP. POWER in particular is like a more fleshed out RISC-V, and they could totally rework the memory/interconnet heavy mainframes to take advantage of the AI boom.

They just... don't.

raverbashing
I wonder what's the competitive advantage of POWER in terms of cost/performance compared with x86

(And sure they have their mainframes, but POWER seems to be the 'middle of the road' solution from them - at least in cost)

Or it's mostly about "this IBM solution only runs under Linux or AIX in POWER6)

brucethemoose2
Their biggest chips are, SMT, memory and interconnect heavy. Seems like they would be quite competitive in memory-bound database loads.

But the big thing with POWER is that its open source. Generally, whatever virtues apply to RISC-V apply to POWER as well.

abudabi123
Do they still have the unique know-it-all to out distance Tesla's Exa Pod application? NASA and IBM working on AI is a sign that good things may come if it is not a white elephant like SLS seems to be expense wise.
Rochus
Quoted from Taulli, T. (2022): "Modern Mainframe Development", O’Reilly Media, Inc, p. 63:

"Every day 200 times more COBOL transactions are performed versus Google searches.

More than 220 billion lines of code are running today, or about 80% of the world’s total.

About 1.5 billion new lines of COBOL are written each year."

IBM sells and maintains much of the "Big Iron" mainframes running COBOL and provides development services.

See e.g. https://en.wikipedia.org/wiki/Mainframe_computer

bob1029
Mainframe business is still a big deal in a lot of places. Banking, finance, insurance, etc. It is a game of inertia & historical baggage, but that also brings with it a stabilizing aspect. Next time you swipe your debit or credit card, you are almost certainly interacting with at least one mainframe. Likely several if the transaction is processed "online".

You might be able to build a more reliable & cheaper system using x86/ARM in a modern datacenter, but you would have a hell of a time getting regulators to approve it. You'd have to make IBM-free finance/insurance your entire business if you wanted to try it out. For those in the industry, it's usually not even worth talking about until you are at the M&A table and someone has a shitload of capital they would like to set on fire.

Starting from zero is not really an option for the types of businesses that are most entrenched in mainframe usage. Part of it is this legacy, but I strongly believe the other part is "valuable lessons learned over many decades". Not rocking the boat used to be something that really bothered me in my 20s and early 30s. Now, I totally get the mindset. Staying on IBM is the most responsible option in many places. It's not impossible to teach GenZ to write COBOL. Failing that, I am certain we will have an LLM that can bridge the gap soon enough.

Rochus
> It is a game of inertia & historical baggage

New is not always better, especially if your company essentially depends on high availabilty and security of their core systems

> more reliable & cheaper system using x86/ARM

The hardware is just a single aspect of reliability and total cost of ownership, not even the most important one.

> It's not impossible to teach GenZ to write COBOL

Not even Java or C++ as it seems.