Rather than argue in the abstract about legacy software quietly running the world, it is more useful to look at three specific, named companies that put their Delphi usage on the record, in their own words, in case studies Embarcadero still publishes today. These are not marketing copy paraphrased into vague claims about productivity. They are attributed quotes from named executives, with real headcounts, line counts, and dollar figures attached, describing broadcast ad-sales software, three-dimensional geological modeling, and a nonprofit CRM and ERP platform, three industries with nothing in common except a Delphi codebase underneath that none of them, after repeated opportunities, chose to rewrite.
WideOrbit: eight developers in 2000, sixty today
WideOrbit is a San Francisco-based broadcast technology company founded in 1999. Its flagship product, WO Traffic, is used by more than 3,000 television and radio stations, including named clients such as Bell Media, DirecTV, Fox Sports, NBC Universal, and Viacom, and the company's separate automation platform runs another 3,200-plus radio stations on top of that. It is not a niche internal tool; it is the software that runs the advertising operations of a meaningful share of North American broadcast media, and it has been built in Delphi since the company's first product shipped.
“In 2000 we created our first product in Delphi, WO Traffic, with eight developers and Microsoft SQL Server. Today we have 60 developers working on Delphi.”
William Offeman, Executive VP of Engineering, WideOrbit
The growth behind that one-line summary is concrete: WideOrbit invests more than $35 million a year in product development and support, and grew from 350 to 550 employees in roughly two years, with its Delphi developer group expanding as a direct part of that growth rather than shrinking as a legacy afterthought. Anders Ohlsson, Technical Product Manager for WO Traffic's radio and digital products, describes the codebase's reach plainly: almost all of WideOrbit's software is written in Delphi, starting from when the company focused only on television, and now extending to its radio, digital, and cable network products as well.
“It's simply mind-blowing that Delphi allows our team to develop for Win32, Win64, OSX, iOS, and Android with the same compiled code, even when it's 20 years old.”
Anders Ohlsson, Technical Product Manager, WO Traffic for Radio & Digital, WideOrbit
This is not a company coasting on old code either. Offeman describes active work revamping mobile solutions and building new ones with FireMonkey for order entry, order approval, and log viewing, plus new REST-style APIs for scaling web services and cross-platform support, and WideOrbit's senior recruiter, Cristina Gutierrez, describes real, current hiring demand: four open Delphi developer positions at the time of the case study, with more expected. Ohlsson's own framing of what that means for an individual career is direct rather than defensive: Delphi, he says, is a great foundation for building a solid career, a claim that is easier to credit next to a team that grew from eight people to sixty over roughly a quarter century rather than shrinking toward zero.
Euclid Technology: a five-person team beating fifty
Euclid Technology built ClearVantage, an integrated CRM, ERP, eCommerce, and web content management platform aimed specifically at trade associations and nonprofits, entirely in Delphi. At the time of its case study, ClearVantage was in use by more than 100 organizations ranging from 15 to 1,000 employees, and its Member Connect module alone was powering roughly 75 member-facing websites collectively accessed by more than a million people a week. The case study itself dates to the Delphi 2007 and Windows Vista era, and Embarcadero still keeps it published today, which is itself a useful data point: it is being offered, years later, as a still-relevant example of what the platform's compatibility story actually delivers, not retired the way a company usually retires an embarrassing old reference.
The specific technical claim in that case study is worth sitting with. Euclid ported the ClearVantage application, more than 480,000 lines of code plus multiple third-party components, from Delphi 5 to Delphi 2007, spanning several major version jumps, and its service-oriented and web architecture, originally built in Delphi 7, required no more than a recompile to keep running on the new version. That is a specific, falsifiable claim about binary and source compatibility across a real multi-version gap, not a general assurance about backward compatibility as a marketing bullet point.
“Using Delphi allows us to compete against development teams that are ten times as large, because we're more productive.”
Charles Vinal, President and CEO, Euclid Technology
Vinal's fuller explanation for why the company picked Delphi in the first place, over Visual Basic, PowerBuilder, and C++, credits "its robust environment and true object-oriented tools," and describes Delphi as having "consistently risen to meet our needs" as Euclid moved from client-server development to extensive web-based services without abandoning its existing code base. The numbers behind the confidence are specific: Euclid runs 25 total employees with a development team of five engineers, competing directly against rival vendors fielding development teams of 50 to 100 people, and reports a 90 percent win rate when its software is evaluated head-to-head against those larger competitors.
RockWare: two million lines, six people
RockWare, based in Golden, Colorado, has spent three decades in geological mapping and modeling software, and two of those decades building on Embarcadero's tools. Its flagship product, RockWorks, builds three-dimensional models of the earth's subsurface for environmental cleanup, hydrogeology, and industrial minerals customers. It did not start that way: after its first ten years, RockWorks had fragmented into 22 separate software products, and RockWare consolidated all of it into a single Delphi codebase specifically to cut costs and improve efficiency for customers. Today that consolidated codebase runs to two million lines of code, maintained by a team of three dedicated programmers and three designer-testers, six people total.
“[Delphi] allows us to put in millions of lines of code and do it in an efficient way without maxing out memory.”
Jim Reed, Director R&D, RockWare
Reed's fuller explanation ties memory management directly to a real distribution constraint most legacy-software writing skips over entirely: RockWare sells globally, and its executables have to run comfortably on modest hardware over slow connections, not just scale up on a data-center machine.
“Because of that, we can have a monster program that does 50 zillion different things, but can be downloaded comfortably in a reasonable amount of time. Given that 65% of our business is outside of North America, executable files have to be small enough so that someone in Sudan can download data over a lousy connection.”
Jim Reed, Director R&D, RockWare
The performance numbers behind RockWare's multithreading investment are specific rather than aspirational: a solid-modeling operation that took 15 days when the company started out now runs in 26 seconds, with the application automatically scaling across whatever processor count a customer's machine has, up to 64 processors on the largest projects, without RockWare's own team having to manage that scaling by hand. Delphi's economics show up directly in RockWare's own pricing too: RockWorks sells for roughly $5,000 per seat against comparable competing tools priced near $70,000, a gap Reed attributes directly to what a six-person team can ship on the platform. The company's newest version is adding another million lines of code on top of the existing two million.
“It has definitely helped us put a lot of our kids through college. It's been very good for us. We couldn't live without it. Every day I still look forward to using Delphi.”
Jim Reed, Director R&D, RockWare, on twenty years of building RockWorks in Delphi
What these three actually have in common
None of these are open-source projects, none of them have a public commit history to point to, and none of them would show up on a chart built from GitHub stars or Stack Overflow tags. That is a real property of this software, but it is not the interesting one here; it is worth looking at the operational and economic pattern instead of the visibility problem. In all three cases, a small core engineering team, five or six people doing the actual programming, has outbuilt and outsold rivals fielding ten to twenty times the headcount, and done it for one to two decades on a single, continuously maintained codebase rather than a sequence of rewrites. Cross-platform delivery from one source tree shows up in every case as a load-bearing requirement, not a nice-to-have: WideOrbit shipping the same compiled logic across Win32, Win64, macOS, iOS, and Android; Euclid recompiling a 480,000-line application across three major Delphi versions instead of re-architecting it; RockWare distributing a two-million-line application to customers on slow connections on five continents. And in the one case where current hiring data is visible at all, WideOrbit's four open Delphi positions, the team is still small relative to its competitors, but it is actively growing, not being quietly wound down while someone plans a rewrite that never gets funded. The industries themselves are also worth noticing precisely because they have nothing to do with each other. Broadcast ad operations, subsurface geological modeling, and nonprofit membership management do not share a customer, a regulator, or a technology vendor relationship of any kind; the only thing connecting WideOrbit, RockWare, and Euclid Technology is that each one independently arrived at the same tool for the same underlying reason, a small team needing to ship a complex, data-heavy, cross-platform application without hiring its way to the headcount a rewrite in a different stack would have demanded. That kind of convergence, three unrelated companies solving three unrelated problems and landing on the same platform, is a harder thing to wave away as coincidence or inertia than a single case study would be on its own.
That is the actual argument for why none of these three got rewritten, and it has nothing to do with nobody noticing them. Each company looked at the alternative, a ground-up rewrite in whatever the current fashionable stack happens to be, and the economics never worked: not when five engineers are already beating fifty, not when a fifteen-day operation already runs in twenty-six seconds, not when a recompile across three major versions already does the job a rewrite would exist to do.
Where this leaves you
- WideOrbit's Delphi team grew from 8 to 60 people over roughly 25 years while the company's overall headcount tripled in the last two alone, which is evidence of small but genuinely ongoing hiring, not a frozen headcount.
- Euclid's 5-engineer team beating rivals with 50 to 100 engineers is a specific, quoted, attributed claim, a 90 percent win rate on head-to-head evaluations, not an abstraction about productivity.
- RockWare's 2-million-line codebase is maintained by six people total, and the company's own numbers, 15 days down to 26 seconds, are the actual argument for keeping multithreaded native compilation rather than funding a rewrite.
- In all three cases the company evaluated a rewrite implicitly or explicitly and kept building on Delphi instead, repeatedly, across one to two decades each.