"From Proprietary Giants to Open-Source Innovators: The Evolution of Chip Design with RISC-V"

In the pre-open-source era (before the 1990s), designing a chip was a highly proprietary, capital-intensive process dominated by large firms like Intel or IBM. Developing a custom microprocessor from scratch could take 2–5 years and cost tens to hundreds of millions of dollars, factoring in design, verification, and fabrication. For instance, the Intel 80486, released in 1989, reportedly cost Intel around $100 million in R&D (adjusted for inflation, roughly $250 million today), with a development timeline of about 4 years—and that was for a company with established tools, expertise, and infrastructure already in place.

Contrast this with today’s open-source culture, exemplified by initiatives like RISC-V. A 2021 study by the University of Michigan and Google’s open-source silicon program showed that using open-source tools (e.g., OpenROAD, SkyWater PDK) and the RISC-V architecture, a small team could design and tape out a functional chip in under 6 months for as little as $10,000–$100,000, depending on complexity and fabrication costs. You got the hook? Great, now let’s move on.

The Basics of RISC and CISC

RISC Overview:

RISC stands for Reduced Instruction Set Computer. I think the name is obvious, but let’s talk more about it. As mentioned in the introduction, tech giants dominated chip design and manufacturing for a long era where the process of creating a chip required intensive efforts from R&D teams, architects, RTL designers, and foundries. The core philosophy of RISC is to use simpler instructions for faster execution and greater efficiency. With RISC, we can do things in their simpler form. Let’s take a quick, though not entirely practical, example to understand the term "simpler instructions." Say we want to have a square root operation in our instruction set. RISC doesn’t deal with this operation directly as a single complex instruction; instead, it relies on software solutions, which form the ecosystem of RISC, to handle this complexity at the software level—not costing an extra penny in hardware, roughly speaking. So, we deal with basic instructions that, in sequence, form an equivalent operation to the square root. Keep this example in mind; we’ll use it when talking about CISC.

To prove the concept of RISC is working, let’s briefly look at the story of ARM. ARM (originally Acorn RISC Machine) began in 1983 as a project within Acorn Computers, a British company struggling to compete with IBM PCs. Engineers Sophie Wilson and Steve Furber designed a simple, power-efficient RISC processor to power Acorn’s next computer, the Archimedes. Facing financial challenges, Acorn partnered with Apple and VLSI Technology in 1990 to spin off ARM as a separate company, Advanced RISC Machines. ARM’s innovation was its business model: it licensed its designs rather than manufacturing chips, enabling widespread adoption. This RISC-based approach paid off—by the 2000s, ARM processors dominated mobile devices, proving RISC’s efficiency could reshape industries.

CISC Overview:

CISC, on the other hand, stands for Complex Instruction Set Computer, which is obvious as well. Its philosophy is exactly the opposite of RISC: it relies on complex instructions, fewer lines of code, and hardware-driven solutions. Now, if we take the same example of the square root operation, we’ll see CISC doing the whole operation within one instruction. This, of course, will take much more time to execute, but remember that computer architecture is all about trade-offs—we lose time but might save power or simplify software, etc.

The Rise of Intel

Intel’s ascent began in 1968 when founders Robert Noyce and Gordon Moore, both semiconductor pioneers, established the company in Silicon Valley. Their vision was to push the boundaries of integrated electronics, and they quickly made their mark with the 1971 release of the Intel 4004, the world’s first commercially available microprocessor. This innovation laid the groundwork for modern computing as we know it. Intel’s big break came in 1981 when IBM chose the Intel 8088 for its first personal computer, cementing the x86 architecture as the industry standard. Through the 1980s and 1990s, under leaders like Andy Grove, Intel dominated the PC market, fueled by innovations like the 80386 and Pentium processors. The "Intel Inside" campaign in 1991 turned the company into a household name, and by 2000, its market value hit $495 billion, driven by the dot-com boom and its near-90% share of the PC processor market.

Latest Struggles

Recently, Intel has faced significant challenges that have eroded its once-dominant position. By March 2025, the company has struggled with manufacturing delays, notably its inability to transition smoothly from 14nm to 10nm nodes (finally mass-produced in 2019 after years of setbacks), allowing competitors like TSMC and Samsung to leap ahead in process technology. AMD has eaten into Intel’s x86 market share, surpassing Intel’s market cap in 2022, while Nvidia’s GPU dominance in AI has left Intel lagging in a critical growth area. Intel missed the mobile revolution—famously declining to power the iPhone in 2007—and has failed to capitalize on the AI boom, passing on an OpenAI investment in 2017. Financially, Intel reported a $16.6 billion loss in 2024, with its stock dropping 60% since 2021. Leadership turmoil peaked with CEO Pat Gelsinger’s exit in December 2024, replaced by Lip-Bu Tan in March 2025, amid skepticism about Intel’s ambitious foundry pivot and its ability to regain process leadership.

What is RISC-V?

Open-source ISA (Instruction Set Architecture):

RISC-V is a freely available, open-standard instruction set architecture based on Reduced Instruction Set Computer (RISC) principles. Unlike proprietary ISAs like x86 (Intel) or ARM, it incurs no licensing fees, enabling anyone to design, modify, and manufacture processors using its specifications. Overseen by RISC-V International, it fosters collaboration across industries and academia, making it a game-changer in democratizing chip design.

Flexible and Extensible Design:

RISC-V features a modular structure with a small, fixed base instruction set (e.g., RV32I for 32-bit integers) and optional extensions (e.g., “M” for multiplication, “F” for floating-point). This allows customization for specific workloads, from low-power IoT devices to high-performance computing systems, without breaking compatibility.

How is it Being Used?

RISC-V’s open nature and adaptability make it a applicable choice across diverse applications:

• Embedded Systems: Widely adopted in microcontrollers and IoT devices for its simplicity and low power consumption. Companies leverage its customizability to optimize for specific tasks (e.g., sensor processing).
• AI and Machine Learning: Its vector extension (RVV 1.0) supports efficient parallel processing, making it ideal for AI accelerators and edge inference devices.
• Automotive: Used in advanced driver-assistance systems (ADAS) and autonomous vehicles, where custom extensions enhance security and real-time performance.
• Software Development: Supported by major operating systems like Linux (since kernel 5.17), Android (added in 2023), and real-time OS like VxWorks, broadening its ecosystem. Usage spans startups to tech giants, with companies tailoring RISC-V cores to their needs, often integrating custom instructions for competitive differentiation.

What is its Potential in Real-World Applications According to Statistics?

RISC-V’s growth and real-world potential are supported by robust statistics:

• Market Penetration: Over 10 billion RISC-V cores have been shipped by early 2025, according to industry estimates from RISC-V International, underscoring its strong presence in embedded systems and IoT markets. Confidence: 90%. Rooted in RISC-V International’s 2022 milestone of 10 billion cores, with steady growth making this a reasonable 2025 figure.
• Cost and Time Savings: In chip design, open-source tools combined with RISC-V can reduce development costs by up to 99% (from millions to thousands of dollars) and time by 90% (from years to months), per a 2021 University of Michigan and Google study, enabling rapid innovation in custom silicon. Confidence: 95%. Validated by academic and industry outcomes (e.g., SkyWater PDK tape-outs), with consistent reporting in RISC-V circles.
• AI and HPC Potential: RISC-V’s open-source model and vector extension (RVV 1.0) position it as a contender in AI accelerators and high-performance computing, offering flexibility to tailor designs for specialized workloads. Confidence: 90%. Supported by technical capabilities (e.g., vector support ratified in 2021) and adoption trends (e.g., NVIDIA’s use in GPUs), though specific market share gains are less quantifiable.

What Products are in the Market Now Based on RISC-V?

RISC-V has spawned a range of commercial products by March 2025:

• SiFive Processors: SiFive, a pioneer, offers cores like the P650 (high-performance, out-of-order) and E-series (embedded). Their HiFive boards (e.g., HiFive Unleashed) are popular for development.
• SpacemiT Key Stone K1: An octa-core 64-bit RISC-V processor (2024) with 4 TOPS NPU and 50 GFLOPS GPU, powering devices like the BPI-F3 single-board computer, LicheePi 3A, and DeepComputing’s DC-ROMA Laptop II.
• NVIDIA: Uses RISC-V controllers in GPUs (e.g., A100) for on-chip management, showcasing its role in hybrid designs.
• Alibaba T-Head: Chips like the TH1520 (used in Sipeed’s LM4A SOM) deliver 2.5 GHz performance, competing with Raspberry Pi 4 in benchmarks.
• Micro Magic: Claims the world’s fastest 64-bit RISC-V core (2020), outperforming Apple M1 in performance-per-watt, though specific products remain niche. These products span consumer (laptops, dev boards), enterprise (storage, servers), and embedded markets, with China’s push (e.g., SpacemiT’s VitalStone V100 server chip) accelerating adoption.

Challenges limiting RISC-V adoption :

Here’s a concise summary of the challenges facing RISC-V:

• Ecosystem Maturity: Incomplete software/tools and limited OS optimization slow adoption beyond embedded systems.
• Performance Optimization: Achieving high-end performance (e.g., AI, HPC) lags behind ARM/x86 due to design and tuning demands.
• Verification Challenges: Lack of centralized validation complicates safety/security certification for critical applications.
• Competition: Established players like ARM and x86 have deeper ecosystems and R&D resources.
• IP/Legal Risks: Open-source nature invites potential patent disputes with proprietary ISAs.
• Talent Shortage: Limited expertise in RISC-V design hampers development speed.

Today, RISC-V takes it further, slashing costs and timelines, empowering small teams, and fueling innovation across IoT, AI, and beyond. Yet, challenges like ecosystem maturity and competition remind us the road ahead isn’t smooth. As of March 2025, with over 10 billion RISC-V cores shipped and products like SiFive’s P650 or SpacemiT’s K1 hitting the market, the question isn’t whether open-source silicon can succeed—it’s how far it will go. The giants once ruled, but the future belongs to the collaborators.

Lets share thought on Calendly: https://calendly.com/mohammadamra00/let-s-chat