On September 10, 2025, Larry Ellison became the richest person in the world, if only for a brief period. His wealth surged after Oracle, the business software company he founded in 1977, reported strong quarterly results that exceeded expectations. Oracle’s shares increased by more than 36% on that day, a spike that saw Ellison’s fortune swell by US$101B. It was, at the time, the largest single-day increase ever recorded.
In case you’re interested what the adoption of data bases looked like for the customers on the ground, the book In the age of the smart machine by Zuboff, there’s many accounts about this. Though she discussed other changes as well.
Oracle databases are still used widely today. As Ellison reminds us so proudly:
‘The Oracle database is used to keep track of basically everything. The information about your banks, your checking balance, your savings balance, is stored in an Oracle database. Your airline reservation is stored in an Oracle database. What books you bought on Amazon is stored in an Oracle database. Your profile on Yahoo! is stored in an Oracle database.’
A triumph, if there ever was one.
is a bit weak, first that quote is obviously very old, and predates Amazon ripping out the Oracle database. But the other thing is that Oracle now is now a database company, they very successfully expanded vertically into applications, which might deserve a story in itself. Then there is their newer pivot to cloud and then AI, which is still in progress, but is what made Larry rich at the beginning of the article, briefly.
‘The Oracle database is used to keep track of basically everything. The information about your banks, your checking balance, your savings balance, is stored in an Oracle database. Your airline reservation is stored in an Oracle database. What books you bought on Amazon is stored in an Oracle database. Your profile on Yahoo! is stored in an Oracle database.’
A triumph, if there ever was one.
Ironically, 2014 is the year that Amazon started a five year effort to rip Oracle out.
A triumph of switching costs, if there ever was one.
This is a great case study and Ellison is a fascinating person. But I think a lot of kept Oracle successful was their acquisitions and investments post DB e.g. peoplesoft, Netsuite, Oracle Cloud, Java, Sun Microsystems etc. They managed to stay current even if the products themselves were terrible or second or third or even 4th in market (cloud for e.g.). They were in all the tech waves in some fashion even if very minor. Though they never really did consumer products hence aren’t talked about the way Microsoft and their peers are.
I think this is the first true Switching Costs business covered in the Commoncog case library. And the thing that I find most interesting about Switching Cost businesses is what it looks like when they’re taking off.
After all, businesses don’t willingly take on a supplier that comes with large switching costs. Hamilton Helmer points out that the best Switching Cost businesses all emerge during a time of … (my words, not his) ‘FOMO’. And of course when someone says that, you want to seek out actual case studies for yourself to see what it looks like, since all the value is in the detail.
This case study demonstrates what a Switching Cost moat looks like when it’s built.
After it’s built, then of course the way to exploit it is to do lots of bolt-on acquisitions. Quoting from my summary of 7 Powers:
An interesting nuance of Switching Costs is that it is a non-exclusive Power type — just like with Counter-Positioning. To use SAP as an example, IBM and Oracle are competitors to SAP in the ERP software space, and they also benefit from high customer retention rates and Switching Costs. So you could say that the enterprise ERP market sits at sort of a stalemate — at least with regard to existing, locked-in customers. The other interesting nuance to this is that you gain no financial benefit if no additional related sales are made to the customer!
The net result is that whenever you have a high Switching Costs environment, you’ll get:
A land-grab for new customers as the market is growing,
And then a switch to a build or buy strategy for integrated add-on products, to sell to your captive install base. This explains why SAP and Oracle become hugely acquisitive in their later years — and why they buy so many adjacent enterprise software products to sell to their customers.
This case covers the land-grab phase of Oracle’s life. It’s super satisfying to me for that reason: now I have a rough idea of how all those other companies subjected themselves to lock-in!
I think AI has the potential for being that FOMO angle and also changing the economics of building a SaaS business enough that there is a reaL chance for some companies to make a dent in high switching cost businesses. I’m seeing some AI native ERPs etc emerging and will be interesting to watch. Also legacy SaaS businesses being scooped up by PE is also a factor. Latest rumors point to workday being a target. Maybe we should make a list of potential targets. I’d put Asana and a few others on that list
We are both trying to say that this needs to be more explicit. The case starts with Larry becoming the richest person in the world, which was nothing to do with the landgrab phase, and you never mention that there were other phases, or make this landgrab framing explicit, or give a clear timescale. The vast majority of people still think that Oracle is a database company.
Good point. I’ll have to think about how to editorialise this within the case itself.
To be fair, though, my read of Oracle’s arc is similar to that of the LVMH arc, or the pattern described in How to Become an Asian Tycoon, and this is a sensemaking frame that is intentionally left to the reader. That is: the vast majority of large diversified businesses have a moat-protected, cash flowing core business, which they may expand outwards from. And it almost doesn’t matter how they expand, so long as that moat-protected core business remains moat-protected and continues to throw off cash. As with many Asian tycoons, the moat protects the company from a lot of bad capital allocation decisions.
So the puzzle in replicating this is two-fold:
How do you get that initial business?
Which moats have the longest duration?
My understanding is that Oracle’s legacy database business remains 21-24% of total revenue at about $20B annually last year, and the switching costs of that install base remains strong. Percentage wise, it is a smaller and smaller part of Oracle’s overall business, but the absolute dollars thrown off is massive. The crossover (that is, when the other business lines overtook the legacy database business) appears to be in 2007, 30 years after its founding.
I am ok with calling Oracle a database company (at least for most of its life), and focusing on the dynamics of that first business.
Edited to add: I still think I need to edit this case though! You’re right in that there should be some kind of editorialising, perhaps at the end, to contextualise the events here. Give me some time; I will probably need to rewrite the entire last section.
Grad: You raise an interesting theory, and we’re going into the 2000s, but the decision
to buy PeopleSoft, which is basically an applications only company, if that wasn’t a big money
maker, and it wasn’t driving the sale of the relational systems, what was the logic of that?
Jacobs: It’s very simple. It created a strategic footprint in our customers. It gave us a
whole stack, a credible stack. And we could now sell at a higher point into the companies, into
the board room. And, our large customers wanted to consider us a strategic partner, rather than
just a vendor of technology. So, it has, actually, had a big impact on the way our sales force
could sell.
At this point these applications were in many cases vertical on top of the database, and this worked well, the document talks about how their competitors did not do this.
I think the interesting point where things change is cloud, which comes after that, and Oracle’s reaction to that.
From that doc, they really loved destroying the competition
And, I think, that period of 1993 to 1997 to 1998 when I was there, it was the period of not just
compete with, but eliminate the other competitors that were the independent relational database guys. And Larry’s philosophy was very clear: you only compete with one company per year. When I took over the database group, it was just the end of eliminating of Ingres. Ingres was
basically dead and bloodied on the floor never to be heard from again, in terms of being a
serious competitor. And they’ve been resurrected as an open source, but not a real threat in
any way. And each year, as I went into Larry’s office for the planning meeting, it was the
question of who is it this year? So, next year, it was Sybase. And you actually get up and give
speeches about, “Well Informix is doing some pretty good stuff,” to take the pressure off, and
every ounce of effort was focused around how do you basically eliminate Sybase? And, it was
not good enough for Oracle to win. It was important for somebody else to lose.
fun fact is that Ingres became Postgres, and is still standing as a database, but not as a company…
Novice take: when the Oracle Version 2 was completed, it was a relational database that achieved breakthrough performance. Yet, the market did not really have enough awareness, or belief, or willingness to learn new behaviors:
Oracle had no choice but to rely on sales revenue to keep operations going. Sales was hard because customers were very sceptical. After all, it was Oracle’s first commercial database product and the earliest database product to use the relational model and SQL. This made adoption trebly difficult: first, customers might not have heard of the relational model. If they had heard of it, they likely did not believe that it worked. Finally, even if they were convinced that it worked, they would have to learn a completely new language to interact with the database.
What puzzled me is that, faced with what seems to be a market timing problem (during the beginning of a technological window), Ellison’s next course of action is to reimplement the whole thing in C so that it becomes portable? It seems very bold, but it’s hard to conceptualize how he reasoned that such a move would resolve the skepticism that customers had. Or did he reason that eventually skepticism would fade away as the technological window opens more fully, and that he should aim for the next bit thing that guarantees Oracle’s moat when that does happen.
I am just imagining if I were in his shoes, and I saw that the market has yet to accept this awesome technology, no way I’d think “Let’s make this technology portable!”
Worth noting: The minicomputer market was fragmented; before IBM entered the market in 1976, DEC had an estimated 40% share of the market. By 1978, IBM was a significant player, Data General was a significant player, and there were many other smaller players. DEC got back to 40% market share with VAX, but that still left the majority of the market out of reach, to say nothing of mainframes
I would be surprised if Ellison’s sales team wasn’t letting him know about issues selling to government departments and enterprises that used multiple platforms, or were IBM or Data General shops
Aye, exactly what @mgoodrum said. The obvious ‘safe’ thing to do is to just rewrite for each platform, again and again, doing sales as they went. Taking on a loan and then spending 5 bloody years doing the rewrite in C, whilst you don’t do any new features … god, that must’ve felt horrible.
There was a reference to the sales strategy – framed here (and commonly) as hunting and farming. Oracle apparently evolved its sales org into a farming-first strategy but started as a hunting-first strategy in the US. I was curious why and how Ellison came to prefer farming over hunting presumably before they had multiple products and services.
Apparently, it had a lot to do with an accounting blow-up in early 1990. From the SEC bulletin (in 1993 at the conclusion of the civil case):
The complaint alleges that Oracle’s materially inaccurate financial reports resulted from an inadequate internal accounting control system that failed to detect double invoicing of customers for products and/or technical support services, invoicing of customers for work that was not performed, failure to credit customers for product returns, booking revenues that were contingent and premature recognition of other revenue. The complaint further alleges that, as a result, Oracle failed to maintain accurate books and records as required under the federal securities laws.
In a hunting strategy, the sales organization is concerned primarily with making a sale, then moving on to the next sale. Farming is focused on nurturing existing customers, with the the goal of up-selling and cross-selling the customer to increase revenue (and obviously, farming was much more suited to Oracle’s evolution as they added more products and services).
But hunting burned them because the accounting failures were at least in part a result of not being close to the customer’s actual ongoing activity…with farming, these errors would have had a higher probability of being prevented.
The accounting errors caused Oracle stock to drop 80% from peak to trough in 1990-1991. A good buying opportunity if you had the view that these accounting issues were temporary and fixable (which they were; it was not fraud) and didn’t actually economically understate the earnings potential of the business. In other words, the stock was down 80% but intrinsic value was not.
Also ironic that the actual accounting issues sound like issues of bad database management!
Apparently Sam Altman has been reading this piece, according to The Information “Altman likened the situation to what Larry Ellison faced with Oracle in the 1980s, when Ellison needed to figure out how to change human behavior and convince businesses to store data with his company.” referring to this podcast episode (havent listened yet) Sam Altman, OpenAI - David Senra
