Loading organizations...
§ Private Profile · San Jose, CA, USA
Enabling customizable hardware with programmable logic
In 1984, Ross Freeman, Bernie Vonderschmitt and James Barnett had the radical idea to build logic devices that users could program themselves and customize for a specific use after they had been manufactured. The three had worked together at Zilog, one of the larger manufacturers of microprocessors at the time. What they saw was an industry happy to manufacture programmed read-only memory (PROM) devices in spite of the limitation that they were hardwired and unchangeable once made. In fairness, there had been little incentive to change. Semiconductor companies were profitable from producing mask-programmable generic circuits. Indeed, the counterargument was that designing multiple circuits for different functions would complicate manufacturing and lower operating margins.
But it was the limitation in the PROM that led Ross to see the opportunity for a new technology:field programmable gate array (FPGA) logic devices:that could be configured by a user after manufacture. After Xilinx shipped its first product in 1985, the company took off like a rocket. The flexibility was just what customers wanted. While FPGAs first took root in telecommunications and networking, they were rapidly deployed in the automotive, consumer and industrial sectors.
Ross, James and Bernie had unsuccessfully pitched Zilog on the FPGA. In 1984, they met with John Doerr who immediately recognized the potential in FPGA, given his prior experience at Intel. We agreed to invest $862,000. John asked Floyd Kvamme, who had just joined Kleiner Perkins, to sit on Xilinx’s board. Floyd was a respected semiconductor veteran who had been on the founding team of National Semiconductor and, prior to joining Kleiner Perkins, had been executive vice president of sales and marketing for Apple. Floyd had the perfect background to help this young company get off the ground.
Xilinx has raised $8.0M across 1 funding round.
Key people at Xilinx.
Xilinx was founded in 1984 by James Barnett (Founder) and Ross Freeman (Founder) and Bernie Voderschmitt (Founder).
Xilinx has raised $8.0M in total across 1 funding round.
Xilinx was founded in 1984 by James Barnett (Founder) and Ross Freeman (Founder) and Bernie Voderschmitt (Founder).
Xilinx has raised $8.0M in total across 1 funding round.
Xilinx's investors include Belmore Capital, National Science Foundation, Xilinx.
Key people at Xilinx.
Xilinx has raised $8.0M across 1 funding round. Most recently, it raised $8.0M Series A in January 2018.
| Date | Round | Lead Investors | Other Investors | Status |
|---|---|---|---|---|
| Jan 3, 2018 | $8M Series A | — | Belmore Capital, National Science Foundation, Xilinx | Announced |
# Xilinx: Pioneering Programmable Logic in Semiconductor Design
Xilinx was a transformative semiconductor company that revolutionized hardware customization through field-programmable gate arrays (FPGAs)—chips that could be reprogrammed after manufacturing to perform different functions.[2] Rather than designing fixed-purpose silicon for every application, Xilinx enabled engineers to deploy flexible, reconfigurable logic that adapted to evolving requirements. The company served a broad ecosystem spanning data centers, telecommunications, automotive, aerospace, and consumer electronics, solving the fundamental problem of hardware inflexibility: the tension between specialized performance and manufacturing economics.
From its 1984 founding through its acquisition by AMD in 2022, Xilinx grew from a Silicon Valley startup into a $1.7 billion revenue enterprise by 2006, employing over 3,000 people globally.[1] The company's impact extended beyond financial metrics—it created an entirely new category of programmable semiconductors that became essential infrastructure for industries requiring rapid iteration and customization without the cost and time of custom chip design.
Xilinx emerged from a moment of technological conviction and entrepreneurial frustration. In 1984, three chip engineers—Ross Freeman, Bernard Vonderschmitt, and James V Barnett II—all working at Zilog Corporation (an Exxon subsidiary specializing in integrated circuits) recognized an untapped market opportunity.[2][3] Freeman attempted to convince Zilog to invest in field-programmable gate arrays, targeting what was then only a $100 million market. When Zilog declined, Freeman and Barnett departed to partner with Vonderschmitt, a seasoned 60-year-old industry veteran who had previously spent 20 years at RCA heading its solid-state division.[3]
The trio raised $4.5 million in venture capital and incorporated Xilinx in February 1984, beginning product sales by November 1985.[1][2] The timing proved prescient. By 1987—just two years after shipping their first product—the company had received over $18 million in venture funding and was generating annualized revenues near $14 million with FPGA chips featuring 9,000 gates that could compete with advanced non-programmable alternatives.[1][3] This rapid traction reflected a fundamental shift in semiconductor economics: the FPGA model made specialized circuits profitable by eliminating the need for expensive custom fabrication runs.
Xilinx pioneered what became known as the "fabless" strategy—designing chips without owning manufacturing facilities.[5] Rather than investing billions in semiconductor foundries, the company formed close partnerships with contract manufacturers, freeing capital for design, support, and marketing. This architectural choice proved transformative, allowing Xilinx to scale rapidly while maintaining asset efficiency.
The fundamental innovation was architectural: FPGAs contained arrays of programmable logic blocks that users could configure post-manufacture through software. This decoupled hardware design from manufacturing timelines, enabling rapid prototyping, customization, and iteration. The company's name itself encoded this philosophy—the "X" represented programmable logic blocks, while "linx" symbolized the programmable connections linking them together.[2]
Xilinx invested heavily in design tools and software that made FPGA programming accessible to broader engineering teams. The company acquired Hier Design, Inc. in 2004, a software firm specializing in FPGA design tools, strengthening its position as an end-to-end platform provider rather than merely a chip vendor.[1] This ecosystem approach created switching costs and developer loyalty.
By 1990, Xilinx had cultivated a customer roster including Apple Computer, IBM, Compaq, Hewlett-Packard, Fujitsu, Sun Microsystems, and Northern Telecom—the computing and telecommunications giants of that era.[1] This blue-chip customer base validated the FPGA model and created network effects as success stories drove broader adoption.
Xilinx occupied a unique position at the intersection of three powerful trends. First, the customization imperative: as computing applications diversified—from telecommunications switching to graphics acceleration to signal processing—the one-size-fits-all processor model became increasingly suboptimal. Xilinx's programmable logic offered a middle path between custom silicon and general-purpose CPUs.
Second, the rise of fabless semiconductor design: Xilinx's success demonstrated that semiconductor innovation could be decoupled from manufacturing, enabling smaller, more agile companies to compete with vertically integrated giants. This model became the dominant paradigm for the industry, spawning companies like Qualcomm, NVIDIA, and AMD itself.
Third, the acceleration of hardware iteration cycles: as software development methodologies embraced rapid iteration and continuous deployment, hardware remained locked in multi-year design cycles. FPGAs compressed this timeline, allowing hardware engineers to adopt more software-like development practices. This became increasingly valuable as edge computing, 5G infrastructure, and AI accelerators demanded rapid customization.
Xilinx's influence rippled through the entire semiconductor ecosystem. By proving that programmable logic could deliver both performance and flexibility, the company legitimized an entirely new category that competitors like Altera (later acquired by Intel) rushed to enter. The FPGA market became a proving ground for emerging applications—machine learning accelerators, network processing, and specialized compute—before those workloads migrated to custom silicon.
Xilinx's trajectory from startup to $1.7 billion enterprise to AMD subsidiary reflects both the company's success and the consolidation pressures reshaping the semiconductor industry. When AMD announced its acquisition in October 2020 (completed February 2022 for approximately $60 billion), it signaled that programmable logic had become too strategically important for independent operation.[2] AMD sought to integrate Xilinx's FPGA capabilities with its CPU and GPU portfolios, creating a more comprehensive heterogeneous computing platform.
The brand itself was phased out in June 2023, with Xilinx products rebranded under AMD.[2] This transition reflects a broader industry pattern: as FPGAs matured from niche components into mainstream infrastructure for data centers, automotive systems, and AI workloads, the competitive advantage shifted from standalone FPGA vendors to companies offering integrated hardware-software stacks.
Looking forward, the programmable logic market will likely be shaped by three forces. AI acceleration will drive demand for customizable compute, as different models and inference patterns benefit from tailored hardware. Edge computing will require flexible logic at the network periphery, where standardized processors prove inefficient. Supply chain resilience will favor companies offering multiple pathways to specialized compute, reducing dependence on any single architecture.
Xilinx's legacy endures not in its brand but in its fundamental insight: that the future of computing belongs to systems that blur the boundary between hardware and software, enabling rapid customization without the cost of custom silicon. In that sense, the company's absorption into AMD represents not an ending but a vindication—programmable logic has become too important to remain a standalone business. The question now is whether AMD can leverage this inheritance to compete with NVIDIA's dominance in AI acceleration and custom silicon, or whether the next generation of specialized compute will emerge from companies yet to be founded.