The Open Hardware Movement: Arduino, Raspberry Pi, and the Democratization of Making

The Problem with Hardware

Software has always had an advantage over hardware when it comes to openness: it can be copied, modified, and redistributed for essentially zero cost. A developer who improves a piece of open-source software can share that improvement with the world by pushing to a repository. The economics are clean.

Hardware does not work this way. A circuit board must be fabricated. Components must be ordered, soldered, and tested. A design file is not a product; it is the blueprint for a product that someone must pay to manufacture. The gap between “open design” and “working device” involves supply chains, minimum order quantities, and capital.

This asymmetry meant that for most of computing’s history, hardware remained proprietary almost by default. The cost and complexity of manufacturing created a natural moat that software simply did not have. Even when companies released specifications — as Intel did with the IBM PC architecture, enabling the clone industry — this was a strategic business decision, not a philosophical commitment to openness.

The open hardware movement that emerged in the early 2000s had to grapple with this asymmetry directly. Its pioneers were not arguing that hardware could be shared with the same frictionlessness as software. They were arguing that sufficiently open hardware designs, combined with the right communities and cheap-enough manufacturing, could achieve something analogous: a world where people without specialized knowledge could build, modify, and share physical computing devices.

The Microcontroller Problem (Before Arduino)

Before 2005, working with microcontrollers — the small, embedded processors that sit inside thermostats, robots, scientific instruments, and countless other devices — required expertise that most creative practitioners did not have and most educational institutions did not teach.

A microcontroller is not a computer in the sense that a laptop is a computer. It has no operating system, no file system, no display. It is a chip with a small amount of memory, a processor running at modest speed, and a set of pins that can be read or written to control external electronics. Programming one required understanding not just code but timing, electrical signals, register manipulation, and the specific quirks of whichever chip family you were using.

The most popular family for hobbyist use in the early 2000s was Atmel’s AVR series. Getting a program onto an AVR chip required a programmer — a hardware device that connected the chip to a computer — and navigating a toolchain of compilers and flashing utilities that had been designed for professional embedded engineers, not artists trying to make interactive sculptures.

The result was a hard barrier between the world of software development, where building something interactive was increasingly approachable, and the world of physical computing, where every project began with weeks of hardware setup before any actual creation could begin.

Massimo Banzi and the Birth of Arduino (2005)

Massimo Banzi was a professor at the Interaction Design Institute Ivrea (IDII) in Ivrea, Italy — a graduate school that brought together designers, artists, and technologists to explore how computing and physical objects could interact. His students were not computer science majors. They were designers who needed electronics to do things and had no patience for the deep technical prerequisites that microcontroller development traditionally demanded.

Banzi had been using a development board called the Wiring platform, created by his student Hernando Barragán, which simplified microcontroller programming substantially. But even Wiring was not simple enough, and it was expensive — around thirty euros per board.

In 2005, Banzi, together with David Cuartielles, Tom Igoe, Gianluca Martino, and David Mellis, developed Arduino. The name came from the Bar di Re Arduino in Ivrea, a pub the team frequented.

Arduino’s design embodied a specific philosophy: make the easiest path also the correct path. The board cost less than thirty euros. It connected to a computer via USB — no separate programmer required. The development environment was a single download. A program to make a light blink was five lines of code. More importantly, those five lines could be written, compiled, and running on the hardware in under a minute.

The choice to release Arduino under open-source hardware licenses was partly philosophical and partly strategic. The IDII was closing due to funding difficulties, and Banzi wanted to ensure that his work would outlive the institution. By making the designs open, he also invited the community to manufacture their own boards — which drove down prices and created a global ecosystem of compatible hardware.

The effect was dramatic. Arduino did not just make microcontroller programming accessible; it created an entirely new category of maker. Artists built interactive installations. Musicians built custom instruments. Scientists built environmental sensors. Teachers built robotics curricula. By 2011, over 300,000 Arduino boards had been sold; by 2015, the number was in the millions. The long tail of Arduino projects documented on Instructables and Hackster.io became, effectively, the world’s largest practical electronics curriculum.

The Raspberry Pi and the Computer for Every Child (2012)

Arduino solved the microcontroller problem. It did not solve the computer problem.

An Arduino is excellent for reading sensors and controlling motors. It is not a computer in the sense of running a web browser, playing video, or running Python. For educational settings that wanted to teach programming in a broader sense — to give children a device they could write code on, experiment with, and break without consequence — Arduino was not the right tool.

This was the problem that Eben Upton and his colleagues at the Raspberry Pi Foundation set out to solve.

Upton was a computer scientist who had worked at Broadcom, the semiconductor company, and had become concerned about the declining interest in computing among applicants to Cambridge University’s computer science program. In the 1980s, students had arrived having programmed on BBC Micros and Sinclair Spectrums — cheap home computers designed to be programmed, not just used. By the 2000s, students arrived having used computers extensively but without having programmed them.

The Raspberry Pi Foundation’s answer was a complete computer — processor, memory, graphics, USB ports, HDMI output — on a credit-card-sized board, sold at a price that made buying one for a child’s bedroom realistic: $35.

The Raspberry Pi Model B launched on February 29, 2012. The foundation had expected to sell perhaps ten thousand units to educators and hobbyists. They sold a hundred thousand on the first day, crashing their distributors’ websites. Within a year, they had sold a million boards. By 2021, the Raspberry Pi Foundation had sold 40 million devices across multiple generations.

The Raspberry Pi ran Linux. It had an HDMI port that connected to any television. It had GPIO (General Purpose Input/Output) pins that, like Arduino, could read sensors and control electronics. It bridged the two worlds: a real computer that could run real software, but one that was also physically accessible and cheap enough to experiment with destructively.

The educational impact was real but incomplete. The foundation’s hope that Raspberry Pis would appear in every school was partially realized — many school districts bought them for computer science classes — but the original vision of the home computer as a learning tool was not fully recaptured. Children who received Raspberry Pis often used them as cheap media centers running Kodi rather than as platforms for learning to program. The device that was designed to teach programming was, for many users, just a very small desktop computer.

This was not a failure; it was a different success than intended. A $35 Linux computer with GPIO pins proved to be exactly what countless projects needed: industrial control systems, environmental monitors, digital signage, network appliances, home automation servers. The Raspberry Pi became the standard platform for a generation of small-scale computing projects that previously required either expensive single-board computers or purpose-built hardware.

The Maker Movement and FabLabs

Arduino and Raspberry Pi were devices; the Maker Movement was a culture.

The term was popularized by Dale Dougherty, founder of Make magazine (launched in 2005), who argued that the combination of cheap electronics, open-source software, and accessible fabrication tools was enabling a new form of manufacturing — not industrial, not hobbyist in the traditional sense, but creative and collaborative in ways that blurred the line between consumer and producer.

The physical manifestation of this culture was the FabLab — Fabrication Laboratory — a concept developed by Neil Gershenfeld at MIT’s Center for Bits and Atoms around 2001. A FabLab was a shared workspace containing a specific set of fabrication tools: a laser cutter, a CNC router, a vinyl cutter, electronics workbenches, and increasingly, 3D printers. The idea was that a collection of tools that collectively cost tens of thousands of dollars could serve dozens of people who individually could not afford them.

The first FabLab outside MIT was established in 2002 in Pabal, India, as part of a rural community fabrication project. By 2012, there were over 100 FabLabs worldwide; by 2020, over 1,800 in 100 countries — a network of physical spaces that made advanced fabrication accessible to communities that would never have had access to it otherwise.

Maker Faire, the exhibition format Dougherty launched in San Mateo in 2006, became the cultural event of the movement: a combination of science fair, craft fair, and engineering showcase where participants shared projects ranging from LED-lit costumes to working robots to custom musical instruments built from salvaged hardware. At its peak, the Bay Area Maker Faire attracted over 100,000 visitors.

The Maker Movement carried a specific ideological charge: that the concentration of manufacturing in large corporations and distant factories was not a permanent feature of industrial modernity but a contingent outcome of cost structures that were now changing. If digital fabrication tools continued to fall in price, the argument went, individuals and small groups would be able to manufacture things that previously required factory-scale investment.

This vision was only partially realized. FabLabs are real and valuable, but they did not transform manufacturing at scale. The cost differential between digital fabrication and factory production remained enormous for high-volume goods. What the Maker Movement produced was not a manufacturing revolution but something perhaps more durable: a global community of people who understood both software and hardware and could make things at small scale.

RepRap and the Self-Replicating Machine (2005)

The tool that most embodied the Maker Movement’s utopian logic was the 3D printer — and specifically, the RepRap project.

Adrian Bowyer, a mechanical engineering lecturer at the University of Bath, had a specific and radical idea: a machine that could print most of its own components. If the machine could reproduce itself, it could spread through a community without the need for a manufacturer — each machine could produce the parts for the next machine, and the design being open-source meant that improvements could propagate across the entire ecosystem.

RepRap — short for Replicating Rapid Prototyper — was launched in 2005 with its first complete machine, the Darwin, functional by 2008. The Darwin used fused deposition modeling (FDM) — the same process used in commercial 3D printers — to deposit melted plastic layer by layer. The printed parts of the Darwin included the frame, many of the mechanical components, and various brackets and fittings. The non-printed parts — motors, electronics, rods, fasteners — were sourced as commodities.

The RepRap project was released entirely under the GNU General Public License. The effect was the same as it had been for Linux and for Arduino: a community explosion. Within months, developers around the world were improving the design, fixing flaws, adding features. The Prusa Mendel, designed by Josef Prusa in 2010, became the most widely replicated variant — and Prusa’s company, Prusa Research, grew from the open-source community into one of the leading commercial 3D printer manufacturers, retaining its commitment to open-source designs while building a sustainable business.

3D printing went from a $200,000 industrial technology accessible only to large manufacturers to a $200 desktop appliance accessible to anyone within a decade. This price collapse was substantially driven by the RepRap project: the open design lowered the cost of entry, the community improved the technology faster than any single company could, and the expiration of key FDM patents in 2009 allowed the commercial market to emerge without licensing barriers.

The self-replication ambition was only partially realized — no 3D printer can print its own motors or electronics — but the cultural impact of framing a machine as a community project rather than a product was profound.

RISC-V: Opening the Instruction Set (2010)

All of the hardware democratization described so far operated at the level of complete devices: boards that users could program, modify, and build upon. Below that level — at the level of the processor itself — hardware remained deeply proprietary.

A processor’s instruction set architecture (ISA) defines the language that software speaks to hardware: the set of operations the processor can execute, how memory is addressed, how interrupts are handled. The two dominant ISAs in computing — x86 (Intel and AMD) and ARM — are proprietary. Intel owns x86 and licenses it tightly; ARM Holdings licenses the ARM ISA on commercial terms that, while not prohibitive, impose costs and restrictions that affect any chip designer.

For a startup or research institution wanting to design a custom processor, the licensing situation was a significant barrier. You could design a chip around an ARM core, but you paid ARM for the privilege. You could design a chip from scratch, but you needed to invent your own ISA, which meant no software ecosystem and no compiler support.

RISC-V was born as an academic project at UC Berkeley in 2010, led by Krste Asanović and David Patterson (who had co-invented the RISC — Reduced Instruction Set Computer — concept in the 1980s). The goal was pragmatic: they needed a clean instruction set for research projects and teaching, and they wanted one they could modify freely and publish without commercial restrictions.

The resulting ISA was designed to be clean, modular, and extensible — but more importantly, it was released as an open standard under a permissive license. Anyone could implement a RISC-V processor without paying royalties. Anyone could extend the ISA for specialized applications without permission.

The timing coincided with the beginning of a broader geopolitical concern about ISA dependency. China’s technology companies and research institutions, confronting export restrictions that threatened their access to ARM and x86 chips, had strategic reasons to invest in an open alternative. Western concerns about chip supply chains, crystallized by the COVID-19 semiconductor shortage and geopolitical tensions with Taiwan, reinforced the argument for sovereign ISA capacity.

By 2022, RISC-V had moved from a research project to a genuine commercial ecosystem. Western Digital had adopted RISC-V for storage controller chips. SiFive had built a startup around RISC-V processor cores. Alibaba’s chip design subsidiary had produced RISC-V processors used in embedded applications. The RISC-V International consortium counted over 3,000 member organizations across 70 countries.

RISC-V will not displace x86 in laptop processors or ARM in smartphones in the near future — the software ecosystem gap is enormous and will take years to close. But it has established that a viable commercial ecosystem can be built around an open ISA, which changes the calculus for future chip design: the cost of owning your own ISA has fallen dramatically.

Dead End: MakerBot and the Limits of Open-Source Hardware Business

MakerBot Industries began in 2009 as one of the most successful embodiments of the open hardware philosophy.

Bre Pettis, Adam Mayer, and Zach “Hoeken” Smith founded MakerBot in Brooklyn as a direct commercial extension of the RepRap community. Their first printer, the Cupcake CNC, was released under open-source licenses in 2009. The business model was RepRap’s model made commercial: sell kits of the non-printable parts, make the design freely available, build a community of users who improve the hardware and create a software ecosystem.

MakerBot grew quickly. The Replicator (2012) was a polished, complete printer that required no assembly and could be bought and used immediately. The company became the public face of the desktop 3D printing revolution: Pettis appeared on the cover of Wired, MakerBot printers appeared in schools, libraries, and offices across the United States.

In June 2013, MakerBot was acquired by Stratasys for approximately $403 million in stock. Stratasys was a large commercial 3D printing company that had operated in the industrial market for decades. The acquisition was intended to give Stratasys a presence in the consumer market.

What happened next was a textbook illustration of the tension between open-source culture and corporate acquisition. The Replicator 2 (2012), MakerBot’s last pre-acquisition design, was released under an open-source license. The Replicator 5th Generation (2014), designed post-acquisition, was not. MakerBot closed its hardware designs.

The community reaction was swift and lasting. MakerBot had been a beneficiary of the RepRap ecosystem — Pettis himself had been a community member before founding the company, and the early MakerBot designs drew directly on community work. The decision to close the designs felt like a betrayal: the company had built its early success on open-source goodwill and then extracted itself from the community obligations that goodwill implied.

The technical consequences were equally severe. The 5th Generation Replicator introduced a new extruder design — the Smart Extruder — that proved unreliable. Because the design was no longer open, the community could not diagnose or fix the problems themselves. MakerBot’s customer support was overwhelmed. Failure rates were high. The brand’s reputation, which had been built on being the friendly, community-oriented face of 3D printing, collapsed over a period of eighteen months.

Meanwhile, the open-source printer market — Prusa Research, LulzBot, and dozens of smaller manufacturers — continued to improve, driven by the same RepRap community dynamics that had made MakerBot possible. By the time MakerBot’s problems were fully evident, Prusa had captured the community’s trust and was producing printers that were technically superior at lower cost.

MakerBot went through multiple rounds of layoffs, eventually reducing its workforce by over 50%. Stratasys wrote down the value of the acquisition significantly. The consumer 3D printing division that was supposed to be Stratasys’s future in mass-market printing became a cautionary tale about the risks of acquiring community-based companies and then violating the community’s foundational norms.

The lesson is not simply that open-source companies should not close their designs. MakerBot’s closure was, in a narrow sense, understandable: Stratasys was a publicly traded company with shareholders expecting returns on a $400 million investment. The lesson is more fundamental: in open hardware, the community is not just a marketing asset or a customer base. It is the source of competitive advantage. When MakerBot closed its designs, it did not merely offend the community. It severed itself from the community’s engineering work — the libraries, the improvements, the bug reports, the user-generated tutorials. The closed product then had to win on corporate R&D alone, against open competitors who had an entire ecosystem working for them. It could not.

The Stakes: Why Open Hardware Matters

The open hardware movement is not primarily about cost — though open designs do reduce costs. It is about who can participate in creating the physical world.

For most of industrial history, manufacturing was the domain of institutions with capital: companies, governments, universities with well-funded labs. The tools required to make complex physical objects were expensive, the knowledge to use them was specialized, and the supply chains to obtain materials were opaque to outsiders.

Arduino, Raspberry Pi, FabLabs, RepRap printers, and RISC-V chip designs together represent an incremental dismantling of these barriers. Not complete dismantling — a FabLab is not a factory, and a Raspberry Pi is not a server cluster — but enough that categories of people who previously could not make things now can.

The consequences are visible across domains. Scientists in developing countries build environmental monitoring equipment from Arduino and Raspberry Pi that would previously have required purchasing expensive proprietary instruments. Artists create interactive installations that would previously have required hiring electrical engineers. Students learn by modifying hardware they own, rather than using sealed appliances they must not touch. Researchers design and tape out RISC-V chips without paying ISA licensing fees.

The structural effect is that hardware, like software before it, is becoming a domain where talent and creativity matter more than access to capital. This is incomplete and imperfect — the barriers to manufacturing at scale remain formidable — but the trend line is clear.

Whether the Maker Movement’s utopian vision of distributed manufacturing was realistic was always secondary to what it achieved culturally: a generation of engineers, designers, and educators who think of hardware as something that can be understood, modified, and shared — not just used.

For the software movements that paralleled open hardware, see The Open Source Revolution and The Rise of Version Control. For the processor architecture that open hardware increasingly runs on, see The Microprocessor Revolution.

---

📚 Sources

• Banzi, Massimo: Getting Started with Arduino, 3rd Edition (2014), Maker Media
• Upton, Eben & Halfacree, Gareth: Raspberry Pi User Guide, 4th Edition (2016), Wiley
• Bowyer, Adrian: “RepRap — The Replicating Rapid Prototyper” — Robotica, Vol. 29, No. 1 (2011)
• Gershenfeld, Neil: Fab: The Coming Revolution on Your Desktop (2005), Basic Books
• Asanović, Krste & Patterson, David A.: “Instruction Sets Should Be Free: The Case For RISC-V” — UC Berkeley Technical Report UCB/EECS-2014-146 (2014)
• Dougherty, Dale: “The Maker Movement” — Innovations: Technology, Governance, Globalization, Vol. 7, No. 3 (2012)
• Moilanen, Jarkko: “Emerging Hackerspaces — Peer-Production Generation” — Open Source Systems: Long-Term Sustainability (2012)
• Weinberg, Michael: “What’s the Deal with Copyright and 3D Printing?” — Public Knowledge white paper (2013)
• RISC-V International: “RISC-V Technical Specifications” — risc-v.org (ongoing)