AI assistance, without expensive design software
Professional chip design depends on commercial design-automation (EDA) licences that cost hundreds of thousands of dollars a year—one reason it has never been part of education. Magellan Silicon students do not buy any of it. Every design is completed in the Magellan Silicon Studio—Magellan Global’s own AI-assisted design-automation platform for young designers, developed with MTRI. The Studio runs in a browser and carries the whole flow: design entry (block-based logic for beginners, hardware description for those ready for it), simulation, verification checks, FPGA programming, the pre-fabrication checks the shuttle requires, a fabrication tracker, and measurement reports when the silicon returns. Underneath, it is built on proven open-source tools, which is why it can be offered at a fraction of the cost of commercial software and to any school or region.
Inside the Studio, an AI design assistant explains, checks and asks why, rather than designing for the student. Every exchange is kept with the project and becomes part of the disclosure statement that competitions and journals require, so the student can show exactly what was their own work. The engineering notebook, the design document and the public project page are generated from the Studio’s own records.
The Studio is deliberately built in stages. The first cohorts use the browser edition described above. The next stage is a visual, game-like design mode in which a young person can assemble and test a chip of their own as naturally as building a world in a sandbox game, and a schools edition that a teacher can run without a semiconductor background. The aim is simple: the cost and difficulty of designing a chip should stop being the reason young people never do it.
The Magellan Silicon Studio: what it is, and what it is built on
The Magellan Silicon Studio is Magellan Global’s AI-assisted design-automation platform for young designers, built by the group’s engineering team in Taipei with MTRI. It runs in a browser and carries the whole flow: design entry (verified blocks for beginners, hardware description for those ready for it), simulation, verification checks, FPGA programming, the pre-fabrication checks each shuttle requires, a fabrication tracker and measurement reports when the silicon returns, plus the engineering notebook, the disclosure statement and the project page.
Underneath, the Studio uses the open-source toolchain the industry’s open-silicon community has proven on thousands of designs: Yosys for synthesis, Verilator for simulation, OpenROAD and LibreLane for layout, Magic and KLayout for design-rule checks, and open process design kits on 130–180 nm processes (IHP’s SG13G2 and SkyWater’s 130 nm through partner shuttles). No commercial EDA licence is involved anywhere a student works, which is what makes the platform affordable for families and schools in any country; commercial sign-off tools enter only if and when the group opens its own education shuttle with a Taiwanese foundry, and never on the student’s side.
The Studio is built in stages. The first cohorts use the browser edition. Next comes a visual, game-like design mode in which younger students assemble and test logic as naturally as building in a sandbox game, with the same path to silicon underneath. Then a schools edition a teacher can run, and an open registry of silicon-verified student designs that later cohorts can build on. The platform is the group’s long-term investment in making chip design something young people anywhere can do, and the research programme above is what tells us whether each stage works.
The Magellan Silicon Studio and its AI assistant
Every design is completed on the Studio, Magellan Global’s AI-assisted design-automation platform for young designers. It runs in a browser and covers design entry, simulation, verification, FPGA programming, pre-fabrication checks, fabrication tracking and measurement reports, so a family never buys design software. The assistant inside it is built by the group with MTRI on the same principles as AKE, the group’s agentic knowledge engine, and it follows three rules. It explains before it generates, so the student understands a construct before using it. It records every exchange, so the disclosure statement is complete and the student can show what was their own. And it asks why: when a student makes a design choice, the assistant prompts them to write down the reason, and those reasons become the engineering notebook.
There is a fourth rule for competition work. The 2027 ISEF rules allow AI as a research resource when it is disclosed, but do not allow generative AI to write the research plan, the abstract, the poster or the citations. The studio therefore keeps the records, provides the structure and points out what is missing, and the student writes those documents themselves.
Beneath the Studio, the toolchain is open source end to end, so no student depends on a commercial licence, and the platform can be offered to schools and in any region at low cost.
Language, blocks or AI: how a young person designs
All three, as three views of the same design. The student sees blocks; underneath is always one Verilog description; the assistant explains, checks and sets questions beside them. This is the structure that Scratch and MakeCode proved for software (an easy start, a high ceiling, and blocks that switch to text both ways); the Studio extends it one layer down, to silicon.
| Layer | Who uses it | What they see | The rule |
|---|---|---|---|
| Intent (language) | Everyone, in the first week | The student describes what they want in their own words. The assistant answers with a block sketch: which modules, roughly how connected. Not code. | The assistant may propose structure; the student assembles it. |
| Blocks (the main workspace) | Ages 12–15, most of the time | Verified modules from the theme kits: counters, comparators, filters, state machines, PWM, UART, a small ALU. Ports are typed (an 8-bit value, a 1-bit signal, an event); a wrong connection turns red. Clock and reset are wired automatically. | Every block carries a fixed piece of Verilog and its own area and timing figures, shown on a “tile meter” as the design grows. |
| Text (Verilog, or a Python hardware language) | Ages 15–17, and anyone who wants to look inside | “Open the block” shows its Verilog; it can be edited, and new blocks can be written. Students who prefer Python use a Python hardware language. | Blocks and text stay synchronised both ways, so nothing learned with blocks has to be unlearned. |
Three things about hardware trip up students who have only programmed, and the Studio is designed around them.
- Time. Software runs a line at a time; in hardware everything moves at once. A “clock” button advances one cycle and every block’s output changes colour together. That one interaction teaches synchrony faster than any explanation.
- State machines. The hardest thing to assemble from blocks, so students do not assemble them: a state-table block takes a table (state, condition, next state, output) and becomes Verilog.
- Limited resources. The tile meter shows how much of the tile’s area and timing budget the design uses. Students discover for themselves that “adding a multiplier breaks the budget”, which is where comparing, trading off and measuring begin.
One design, three destinations: “Simulate” runs it in the browser; “Program” loads it onto the Silicon Board and the student’s robot or sensor; “Submit” places it as a tile on the cohort’s shared chip; after the silicon returns, “Measure” puts chip and simulation on one chart. The student learns one screen from the first block to the finished work.
The Magellan Silicon Studio, in stages
Now. The browser edition: design entry, simulation, verification, FPGA programming, pre-fabrication checks, the AI assistant, the engineering notebook, a fabrication tracker, measurement reports and the project page. Nothing to install; a family needs a laptop and the development board we ship.
Next. A visual, game-like design mode in which younger students assemble and test logic as naturally as building in a sandbox game, with the same path to silicon underneath. Designing a chip should feel approachable before it feels difficult.
Then. A schools edition a teacher can run with mentor support, and an open registry of silicon-verified student designs that later cohorts can build on. The Studio is developed by Magellan Global with MTRI and the group’s Taipei engineering team, and it is the group’s long-term investment in making chip design something young people anywhere can do.
The standard products
The Studio and the open-source toolchain beneath it are the platform; the products are what a family or a school actually receives. They are deliberately few.
| Product | What it is | For whom |
|---|---|---|
| Silicon Board | One general development board: a small FPGA with a fully open toolchain, a socket for the returned chip, sensor connectors and USB. Programmed from the browser. | Every student; the only hardware platform, used from week one to the returned chip |
| Theme kits (six) | Robot sensing; environment and low power; sound and music; games and graphics; security and identity; streams and finance. Each holds sensors, five verified block templates and example projects. | Ages 12–17, chosen with the mentor to fit the student’s question |
| Magellan Silicon Studio | The browser platform: blocks, text, simulation, FPGA programming, pre-fabrication checks, the AI assistant, notebook, disclosure statement and project page. Free to try; programming and fabrication come with a cohort place. | Anyone; the entry point and the community |
| Platform chip (planned) | A fixed, configurable chip holding the kits’ parameterised modules and a programmable interconnect; a student’s block design becomes a configuration loaded by the board. Lets a student make a hundred products from one design. | The products pack; schools and competition teams |
| Schools edition | Boards, teacher handbook, coach support and a fabrication allocation. | Schools whose teachers have no semiconductor background |
Taiwan: where the chip is made, and the people who make it
Fabrication uses multi-project wafers on mature 130–180 nm processes, through partner shuttle services in the United States and Germany, with packaging, test boards and measurement handled by the group’s engineering team in Taipei. Every cohort is booked on two independent fabrication routes so that a delay on one does not stop a student’s chip.
Taiwan is where the modern semiconductor industry is concentrated, and understanding how chips are actually made—how a shuttle is scheduled, what packaging and test involve, how design services, foundries and assembly plants divide the work—is part of what students take away. The optional Taipei silicon week brings students to that ecosystem in person: visits to fabrication and packaging facilities, measurement sessions in the lab, and time with the engineers who reviewed their designs. In the age of AI, chips are the physical foundation of everything else; Magellan Silicon lets young people see that foundation for themselves.