The RISC-V architecture is taking a significant step forward in the conversation about data centers for Artificial Intelligence. Not because it will suddenly displace x86 or Arm, but because it is beginning to appear where it was previously rarely considered: at the core of large-scale AI computing. This shift in tone has become more evident following SiFive‘s recent funding round, backed by NVIDIA and other investors, and their partnership to integrate NVLink Fusion into future CPU solutions for data centers. The message is clear: RISC-V aims to move beyond being just an open and flexible promise to becoming a real alternative within AI infrastructure.
On April 9, 2026, SiFive announced a funding round of $400 million, valuing the company at $3.65 billion. The company explained that this capital will be used to accelerate its roadmap for high-performance CPUs aimed at data centers. Additionally, CEO Patrick Little indicated that this could be the last major round before a future IPO, though no exact timeline was provided. More than a financial move, this signals an industrial commitment: SiFive wants to leverage the current uncertainty around traditional architectures to offer an open-standard-based alternative.
This context matters. SiFive does not manufacture finished chips for direct sale, but instead markets intellectual property and foundational designs that customers can customize to create their own processors. For years, this space has been dominated by Arm, but Arm’s decision to develop its own chips has shifted part of the balance with its traditional clients. According to Patrick Little, this shift has created an opportunity for more companies to view RISC-V as a pathway offering greater control, customization, and less reliance on a single vendor.
RISC-V no longer wants to stay only at the edge
Until recently, RISC-V was mainly associated with microcontrollers, embedded systems, edge computing, and low-power projects. The new development is that it’s now gaining visibility in the most demanding market segment: AI-focused data centers. SiFive had already hinted at this shift in September 2025, when it introduced new RISC-V IP combining scalar, vector, and matrix compute to accelerate AI from the edge to the data center. However, a true turning point arrived in January 2026, with the announcement of integrating NVIDIA NVLink Fusion into its upcoming data center-class solutions.
The rationale behind this partnership is quite straightforward. AI data centers are no longer designed around general-purpose CPUs and standard networks alone. The bottleneck lies in how CPUs, GPUs, memory, and accelerators connect, and how to do so with improved performance per watt. SiFive and NVIDIA propose that a highly customizable RISC-V CPU, coherently connected and with high bandwidth to NVIDIA GPUs via NVLink Fusion, can provide a more flexible platform for next-generation heterogeneous architectures. This is no small bet: it means bringing RISC-V into the decision-making arena that shapes the future of AI infrastructure.
Furthermore, SiFive argues that its approach makes sense now, as data center operators seek more open and customizable platforms. The company offers wide, out-of-order cores, coherent scalable fabrics, and advanced memory hierarchies designed to better suit specific workloads. In other words, it’s not just ideological openness; the idea is that RISC-V’s modularity can translate into more finely tuned AI infrastructures. At the core, this is the industrial proposition SiFive is offering.
NVIDIA validates a move that once seemed distant
NVIDIA’s support is especially significant. The company is not only an investor in SiFive’s funding round but also involved technically through NVLink Fusion. NVIDIA CEO Jensen Huang stated in January that this integration provides flexibility to combine customizable RISC-V CPUs with NVIDIA accelerators, enabling more scalable and efficient AI infrastructure. For SiFive, this validation adds visibility and credibility in a market where NVIDIA largely determines the pace of AI innovation.
It does not imply that RISC-V will replace dominant architectures in the short term. x86 still has a large installed base in data centers, and Arm has been consolidating its position in hyperscalers and custom chip designs for years. However, it signals that the conversation is opening. Reuters even pointed out that the data center CPU market is heating up, with Arm entering this space more directly, NVIDIA strengthening its position, and Intel responding to a demand that has sometimes exceeded its capacity. RISC-V aims to capitalize on this ongoing reorganization.
It’s also noteworthy that SiFive enters this phase with a somewhat stronger position than often perceived. The company claims to have achieved more than 500 designs using their IP and shipped over 10 billion cores to date. These figures don’t yet render it a dominant player in data centers, but they prove that RISC-V is no longer just an academic curiosity or marginal technology. It has a commercial base, an ecosystem, and now aims for a higher tier.
The key question: will Europe and the market embrace a more open architecture?
The appeal of RISC-V goes beyond the technology itself; it lies in its governance. Unlike Arm, which is controlled by a single company, RISC-V is open and community-driven. This can be particularly attractive now, as major infrastructure buyers seek greater flexibility, customization options, and less dependency on closed ecosystems. SiFive is leveraging this for commercial gain, and NVIDIA seems willing to reinforce this narrative if it broadens its own acceleration ecosystem’s reach.
Nevertheless, the real challenge now is convincing highly demanding clients, demonstrating mature software, integrating with complex toolchains, and proving that open architectures can coexist with reliability, performance, and large-scale support. SiFive has secured funding, symbolic backing, and a favorable narrative, but the next step is proving that it can achieve actual deployments in the field.
Frequently Asked Questions
What is SiFive, and why is it making news in 2026?
SiFive is a company specializing in intellectual property for RISC-V-based processors. It is making headlines because it has secured $400 million in funding, supported by NVIDIA, to accelerate its high-performance CPU roadmap for data centers.
What does NVIDIA’s backing of SiFive entail?
It involves both a direct investment in SiFive’s latest funding round and technological collaboration to integrate NVLink Fusion into SiFive’s future RISC-V solutions. This strengthens the idea of combining RISC-V CPUs with NVIDIA GPUs for AI infrastructure.
Can RISC-V truly compete with x86 and Arm in data centers?
Currently, it lags behind in deployment, but SiFive is trying to leverage the demand for more open and customizable architectures to gain ground in AI. It’s not an immediate replacement but is positioning itself as a viable alternative.
What is the main advantage RISC-V offers for AI compared to other architectures?
Its primary strength lies in openness and customization. SiFive advocates that this flexibility allows designing CPUs more tailored to specific workloads and connecting them more efficiently with accelerators via high-bandwidth interconnects like NVLink Fusion.