It was ridiculous. They seriously wanted to rewrite everything in Java, including office and web browsers. It was 10 times worse than the recent “rewrite in Rust” mania and way more unrealistic.
There's another perspective. Many people were looking for something like Java well before it was released: VM-based, portable, modern object-orientation features, etc.
Case in point: databases. In the early 1990s I worked on a project at Sybase that attempted to rewrite SQL Server from the ground up to bring in object [relational] support. The team focused on VM-based languages as a foundation, which were an area of active academic research at the time. Built-in object support, portability, and ability to support code generation for queries were among the attractions. The project started with Smalltalk (slow!), then moved to an acquired VM technology (it was bad!), and finally a VM we designed and built ourselves. These gyrations were a key reason why the project failed, though not the only one.
When Java came out in 1995--I got access to the alpha release in September--it met virtually every requirement we were trying to fulfill. At that point most attempts to build new databases on other VM tech became instantly obsolete. (Other vendors were also looking at VMs as well.)
Not coincidentally Nat Wyatt and Howard Torf, a couple of key engineers from our project, founded a start-up called Cloudscape to pursue the Java route. They created the database we know today as Derby.
Somewhat more coincidentally, Java became dominant in American DBMS development after 2000. Examples including Hadoop, Druid, Pinot, HBase, etc., are just a few of the examples. I say "somewhat more concidentally" because at that point most of us saw Java as simply more productive than C/C++ alternatives for building reliable, high-performance, distributed systems. That view has obviously evolved over time, but between JIT, dev tooling, and libraries it was definitely true in the early 2000s. It helps to remember how difficult C++ was to use at that time to understand this perspective.
In summary, a lot of the hype was the usual new technology craziness, but Java also met the needs of a population of people that went far beyond databases. There was a basis for our excitement. Just my $0.02.
Edit: typo
The hype accomplished something that would be otherwise impossible: it established Java as a language in what was likely record time. Consider another popular language: Python. It was created about 5 years earlier, yet it rose to prominence about a decade later. Or consider Rust. It is, in many respects, as significant as Java. While major developers were shipping significant applications written in Java within 5 years, Rust is only creeping into important software a decade after its introduction.
The second point is its easy to underestimate the dominance of Microsoft in those days. You think that Microsoft is dominant today, well, that's nothing compared to the late 1990's. Microsoft's market share was closer to 95% for the PC market. The workstation market was starting to crumble due to competition from Microsoft and Intel. About the only thing that was safe were mainframes, and that was highly dependent upon one's definition of safe. Nearly everyone who was competing against Microsoft wanted to see a chunk taken out of them, which meant pretty much everyone since Microsoft had its fingers in so many markets. And, as it turns out, nearly everything did have to be rewritten. Sometimes it was to deliver SaaS over the Internet and sometimes it was to target mobile devices.
> The second point is its easy to underestimate the dominance of Microsoft in those days. You think that Microsoft is dominant today, well, that's nothing compared to the late 1990's. Microsoft's market share was closer to 95% for the PC market.
By the late 1990s Linux had become a viable platform for a whole lot of things, and people were beginning to take notice. Most obviously, that probably put a big dent into the adoption of Windows NT as a server OS on x86 machines, which had been progressing quite well until the mid 1990s. That also probably helped Java because it meant you could seamlessly run your server workloads on "toy" x86 machines or on more "serious" platforms, without changing anything else.
RIIJ was justified too, because people believed the web will end up being java applets all the way down.
I think Java helped in the mainstreaming of memory safe and GC languages, especially in more corporate spaces where C/C++ was still mostly still a thing.
Certainly sysadmins were using a lot of Perl during that time, but for "real" enterprise software, I don't think dynamic-ish languages were as accepted. The use of Perl and rise of Python widened the Overton window.
I predict we will be having buffer overrun CVEs in C/C++ code for as long as we have C/C++ code.
The realities of writing safe, multithreaded C/C++ on processors with out-of-order processing, context switching and branch prediction is simply too complex to get right 100% of the time. Rust makes writing certain code difficult because it is difficult to do/ C/C++ fools you into believing something is safe because you've never encountered the circumstances where it isn't.
We have tools like valgrind to try and identify such issues. They're certainly useful. But you'll be constantly chasing rabbits.
I've seen thread and memory bugs in production code written by smart, highly-paid engineers at big tech companies that have lain dormant for the better part of a decade.
That's why Rust exists.
Rust will never take over and something like C/C++ will always be with us.
It's important, we got it years ago. With Java.
Only applications and web people did believe that particular part of the hype, because they didn't know what it took. And IBM consultants believed it because IBM always managed to sell some weird CPU extension for big dollars that did weird stuff, like their "decode XML extension".