⏱️ Lectura: 11 min
Bryan Cantrill, cofounder of Oxide Computer Company and former distinguished engineer at Sun Microsystems, published an essay on September 20, 2026 that answers a question nostalgia for Sun almost never dares to ask: what did a company that invented Java, popularized Unix in the nineties, and became synonymous with silicon innovation get wrong?
📑 En este artículo
The piece, titled “What Sun got wrong,” came out of an almost trivial trigger: the Oxide team was preparing tribute t-shirts for defunct computing companies for its annual meetup, OxCon, and one of them paid homage to Sun. That t-shirt sparked a reflection Cantrill had already sketched out in 2011 and that now, fifteen years later, he sums up in a single sentence.
TL;DR
- Bryan Cantrill published an essay titled “What Sun got wrong” on his personal blog on September 20, 2026.
- His central thesis: Sun Microsystems didn’t fail because of strategy, it failed because it got bored of running its own business.
- In 2005, a startup running its entire infrastructure on OpenSolaris couldn’t get Sun to sell it hardware over the phone.
- That same startup bought the servers from Dell that same night, helped by an account executive named Steve.
- Scott McNealy ran Sun for 22 years and said he never had to hide his kids’ newspaper out of embarrassment.
- Oracle ended up acquiring Sun Microsystems in 2010, after years of operational decline.
- Cantrill and Steve, the former Dell salesman, went on to cofound Oxide Computer Company.
- The essay grew out of a tribute t-shirt honoring Sun that the Oxide team designed for its annual OxCon meetup.
What happened
Cantrill writes from Emeryville, California, where Oxide’s entire remote team is gathering this week for OxCon, its annual in-person meetup. As part of the preparations, the company’s designer, Ben Leonard, created a series of t-shirts honoring defunct computers and companies. Among them, one dedicated to Sun Microsystems turned out, in Cantrill’s own words, to be “too hot to leave as a surprise.”
The t-shirt reignited a nostalgia Cantrill considers largely deserved. Oxide’s stated mission, in fact, comes directly from the epitaph Scott McNealy wrote for Sun, and from a reflection of his own: across 28 years running the company, McNealy said he never had to “hide the embarrassing newspaper from my kids.” For Cantrill, that line remains a standard every company should aim for.
But nostalgia, he warns, can become suffocating to the point of silencing the uncomfortable question: what did Sun get wrong? Cantrill answers that yes, Sun got plenty wrong, and that he himself, without ever feeling ashamed of having worked there, felt embarrassed more than once by specific company decisions.
Background and history of Sun Microsystems
For much of the nineties and early two-thousands, Sun Microsystems was one of the most influential companies in computing: its SPARC-powered workstations, its Solaris operating system, and its Java language defined a large part of that era’s internet infrastructure. McNealy ran it as CEO for 22 years, from 1984 to 2006, and under his leadership Sun reached billions of dollars in annual revenue.
Cantrill’s essay doesn’t dispute that legacy. What it disputes is an operational pattern that, in his reading, became structural over time: Sun stopped paying attention to the everyday mechanics of selling, delivering, and serving its customers. Cantrill sums it up in a line that anchors the whole essay: “Sun had gotten bored with the mechanics of running a business.”
The episode Cantrill uses to illustrate that diagnosis happened in 2005. A fast-growing startup was running its entire infrastructure on OpenSolaris and wanted to buy a significant number of Sun servers. For Sun, this should have been total validation of its bet on open-sourcing Solaris: the model was working, the customer was growing, and it wanted to pay for its own hardware.
But the customer couldn’t buy. The startup called Sun repeatedly and couldn’t get anyone to call back; when someone finally did answer, they tried to sell the wrong product. The contrast with Dell was stark: that same startup filled out a web form in the middle of the night and, by the next morning, an account executive named Steve was already calling. In under two weeks, Dell had installed the servers in the data center and financed the whole deal with no personal guarantees, based solely on the company’s finances.
Technical details: from OpenSolaris to Oxide
The technical irony is twofold. Sun had open-sourced Solaris in 2005 specifically to attract companies like that startup, which needed a reliable operating system for demanding workloads without paying for traditional proprietary licenses. The strategic bet was correct: OpenSolaris ended up being the foundation on which core ideas of what we now call cloud computing were built. Cantrill himself was one of the engineers who developed DTrace at Sun, the dynamic kernel instrumentation tool still used today to debug production systems without restarting them.
The problem wasn’t the technology, it was the layer around it: sales, support, delivery logistics, customer service. That operational layer is what Dell mastered with almost industrial precision in 2005, and what Sun had stopped tending to. Cantrill closes the anecdote with a detail that changed his professional life: after leaving Sun, he joined that exact startup, the one that never got a response from Sun. And Dell’s Steve ended up getting hired by that same company years later too.
Out of that shared history, years later, came Oxide Computer Company, founded by Cantrill together with Steve. The company builds full-rack servers with an approach deliberately opposite to what Cantrill criticizes in Sun: open firmware, no proprietary BMC (Baseboard Management Controller), which has been a source of critical vulnerabilities in commercial servers, and its own embedded operating system called Hubris, written in Rust and released as open source.
# Clone and build Hubris (Linux, macOS, and Windows with PowerShell)
git clone https://github.com/oxidecomputer/hubris.git
cd hubris
cargo xtask dist app/demo-stm32h7-nucleo/app.tomlThe command above builds a Hubris demo image for an STM32H7 development board; it’s the same workflow, with the same Rust tooling (cargo, rustup), that Oxide’s engineers use to build the firmware that runs on their own servers. A typical manifest file in that repository defines each system task as an isolated unit with its own memory budget:
[tasks.thermal]
name = "task-thermal"
priority = 5
max-sizes = { flash = 32768, ram = 8192 }
start = trueEach Hubris task, like the thermal management one in the example, runs isolated from the rest: if it fails, the system restarts it without affecting the others. It’s a design philosophy different from Solaris in 2005, but connected by the same throughline: build systems the company itself understands and operates end to end, without depending on third parties for critical pieces.
💭 Key point: Cantrill isn’t saying Sun’s technical strategy was wrong. OpenSolaris was the right bet. What failed was the operational execution around that strategy.
How to explore the essay and Oxide’s code
Cantrill’s full essay is published on his personal blog, bcantrill.dtrace.org, and requires no signup or subscription. For anyone who also wants to check the original 2011 reasoning that Cantrill says he still holds fifteen years later, that comment circulated at the time on Hacker News.
Anyone who wants to go a step further and see in code how Oxide translates that “end-to-end ownership” philosophy into a real product can clone the Hubris repository and check the app/ folder, where each Oxide server (known internally as a “Gimlet”) has its own configuration file with the tasks that run on the hardware:
:: Windows (PowerShell), macOS, and Linux all use the same command
git clone https://github.com/oxidecomputer/hubris.git
cd hubris
cargo build --releaseTo confirm the build environment is set up correctly, just check that cargo xtask lists the available applications without missing-dependency errors; the repository itself documents the Rust toolchain requirements for each target board.
Impact and analysis
The lesson Cantrill draws isn’t exclusive to Sun or to hardware. For development teams and tech startups in Latin America, the point is transferable: superior technology guarantees nothing if the operational layer around it, customer care, support, delivery logistics, is neglected. It’s a particularly relevant risk for fast-growing companies that start treating customer service as a cost to minimize rather than a core business function.
One clarification is worth making: this is the perspective of a single former employee, and Cantrill explicitly acknowledges it in his own essay, saying everyone who worked at Sun will have their own take on what went wrong. Nor does flawless operational execution guarantee survival: Oxide competes in a capital-intensive server hardware market, where giants like Dell, HPE, and the major cloud providers operate at far greater scale. Operational discipline is a necessary condition, not a sufficient one.
Sun kept operating for a few more years after the 2005 episode, and Cantrill says he and the rest of the team “tried everything” to right the course. It wasn’t enough: Oracle ended up acquiring Sun Microsystems in 2010, closing the independent chapter of one of the most influential companies in computing history.
⚠️ Note: Cantrill himself points out that this account is his personal interpretation, shaped over years inside Sun; other former employees in different roles may have a different diagnosis of what went wrong.
What’s next
For now, the essay serves as a symbolic close to a story that began with an unanswered phone call in 2005: Cantrill and the Dell salesman who did answer, Steve, now run Oxide Computer Company together, applying in every operational decision the lesson they drew from watching, from the inside and the outside, the company they admire most fail.
Oxide’s full team is in Emeryville this week for OxCon, and the company is expected to keep publishing similar reflections on the computing companies it honors on its t-shirts: according to Cantrill, the point isn’t just nostalgia, but learning from what those companies got right as much as from what they got wrong.
📖 Summary on Telegram: See summary
If you want to get a firsthand sense of the tone and detail of the essay, the most direct route is to open Cantrill’s original post on bcantrill.dtrace.org and read it in full today.
Frequently Asked Questions
Who is Bryan Cantrill?
He’s cofounder and CTO of Oxide Computer Company, creator of the DTrace instrumentation tool, and a former distinguished engineer at Sun Microsystems, where he worked before Oracle acquired the company.
What does Scott McNealy’s epitaph for Sun mean?
McNealy ran Sun for 22 years and used to say that in all that time he never had to hide the embarrassing newspaper from his kids, referring to running the company ethically and without scandals.
Why couldn’t Sun sell hardware to the startup in 2005?
According to Cantrill’s account, the customer couldn’t get anyone at Sun to return its calls, and when someone finally helped, they tried to sell the wrong product, even though the startup was already running its entire infrastructure on OpenSolaris.
What is OxCon?
It’s Oxide Computer Company’s annual in-person meetup, a company that operates remotely the rest of the year and brings its whole team together once a year, as happened this week in Emeryville, California.
How does this story connect to Oxide’s founding?
The Dell executive who helped the startup in 2005, identified as Steve, ended up working at that same company years later alongside Cantrill, and the two went on to cofound Oxide Computer Company.
Where can I read the full essay?
It’s published, in English and freely accessible, on Bryan Cantrill’s personal blog: bcantrill.dtrace.org.
References
- bcantrill.dtrace.org: Bryan Cantrill’s original essay, “What Sun got wrong,” published on September 20, 2026.
- github.com/oxidecomputer/hubris: repository for the open-source real-time operating system Oxide uses on its servers.
- en.wikipedia.org: full history and timeline of Sun Microsystems, from its founding to its acquisition by Oracle.
- en.wikipedia.org: profile of Oxide Computer Company, cofounded by Bryan Cantrill.
📱 Like this content? Follow @programacion on Telegram for daily tech content in Spanish: quick summaries, fresh content every day. @programacion
0 Comments