A rare form of access to Intel technology

Intel has reportedly provided RosaicLabs, a recently incorporated chip startup, with access to processor technology associated with its Atom line. The reported arrangement is notable not merely because it involves x86, but because the technology is said to include register-transfer level, or RTL, code: a detailed hardware description used to turn a processor design into physical silicon.

That differs materially from the conventional sale of a finished processor. With a packaged chip, the customer chooses Intel’s product largely as designed. RTL access can allow a specialist chip developer to incorporate a processor core into a broader system-on-chip, surrounding it with application-specific accelerators, memory interfaces, security functions and input/output logic. The scope of RosaicLabs’ rights, the particular Atom design involved, commercial restrictions and manufacturing plans have not been publicly detailed.

The reporting should therefore be treated carefully. There is no public product roadmap from RosaicLabs, and there has been no detailed public disclosure of the transaction’s contractual terms. But the reported transfer is meaningful precisely because Intel has generally kept control of the implementation of its x86 processors, even while licensing software, patents and other forms of intellectual property across the industry.

Why RTL matters

RTL is the stage at which engineers describe how digital logic operates, including the movement of data between registers and the behaviour of execution units. It is not equivalent to handing over a completed chip layout, but it is far more useful to a semiconductor designer than permission to use an instruction-set label or to run software on existing Intel processors.

A company with suitable engineering resources could use such code as a starting point for a custom chip. Potential applications could include embedded devices, industrial systems, networking equipment or workload-specific edge computing. It could also, in principle, be combined with other intellectual property and produced through a foundry service.

That potential does not establish RosaicLabs’ intended market. Atom has covered multiple generations and use cases over its history, and the precise vintage and capabilities of the reported technology remain unknown. It would be premature to assume that the startup is building a direct competitor to mainstream PC or server processors, or that it has access to every modern x86 feature.

Still, providing reusable processor implementation technology would represent a more flexible model than selling a standard chip. It resembles, in limited form, the approach used by merchant intellectual-property suppliers that license processor building blocks to customers developing their own silicon.

A historical precedent, but a different context

Intel has previously sought to broaden Atom’s reach. In 2009, it announced an agreement with TSMC to port Atom CPU cores to TSMC’s technology platform for system-on-chip customers. That initiative was intended to support customised embedded implementations, showing that Intel has long recognised the value of Atom technology beyond conventional PC processors.

The RosaicLabs report appears different in emphasis. The earlier Intel-TSMC announcement was a public collaboration built around a manufacturing-platform relationship. The newer arrangement concerns a startup with little public operating history and, according to reporting, access to Atom-related RTL. The distinction matters because processor RTL is among the most commercially sensitive elements of a semiconductor company’s design capability.

Intel’s wider business position helps explain why even a narrowly scoped deal attracts attention. The company is trying to develop Intel Foundry into an external customer business, moving beyond a manufacturing organisation built principally for Intel’s own products. Its recent filings describe design enablement, chiplet integration, advanced packaging and foundation intellectual property as parts of that offering. Intel has also said it wants to extend the reach of its core x86 IP through custom silicon and design services.

Seen through that lens, the RosaicLabs arrangement could be an experiment in whether selected processor technology can help attract design work, manufacturing demand or both. It may be a one-off transaction, rather than the beginning of a broad CPU-core licensing programme. But it is consistent with a company looking for ways to derive more value from accumulated design assets while making its foundry platform more relevant to external customers.

Governance questions are unavoidable

The deal has also drawn scrutiny because RosaicLabs’ reported chief executive, Amarjit Gill, has longstanding professional and investment connections with Intel chief executive Lip-Bu Tan. Reporting on the transaction said Gill and Tan had both invested at an early stage in Nuvia, the processor startup later acquired by Qualcomm, and that they had connections through Rivos.

Such relationships are not, by themselves, evidence of improper conduct. Semiconductor development is a specialised field in which investors, executives and engineers often work together across several ventures over many years. Intel appointed Tan as chief executive in March 2025 in part because of his industry experience and extensive ecosystem relationships.

Nevertheless, the combination of a little-known startup, a strategically important form of design access and close professional links makes process important. For Intel, the central governance issue is whether the transaction was evaluated under robust controls, with appropriate treatment of conflicts, intellectual-property protection and commercial value. Public information does not provide enough detail to judge those questions conclusively.

The reputational dimension may matter as much as the legal one. Intel is attempting to persuade external customers that it can be a trusted manufacturing and design partner, while also protecting the proprietary technology that underpins its products. A poorly explained agreement could create doubts about consistency or preferential access. A clearly governed and commercially disciplined one could instead demonstrate a willingness to work with customers in more tailored ways.

The test is execution, not the announcement

The immediate business impact is likely to be limited. RosaicLabs is early-stage, and chip development requires substantial capital, engineering talent, design tools, verification, software work and access to manufacturing. A license alone does not produce a commercially viable processor.

The longer-term significance rests on three unanswered questions. First, can RosaicLabs convert the reported access into a differentiated product and secure funding for its development? Second, will Intel receive manufacturing, packaging, design-services or licensing revenue that justifies the strategic and governance complexity? Third, does Intel repeat the model with other customers?

Intel’s own disclosures underline the urgency of building a credible external foundry business. It has said that it has had few external foundry customers to date and that securing significant customers is important to the economics of future process technologies. A successful custom-silicon engagement would not resolve that challenge on its own. It could, however, offer evidence that Intel can combine its design heritage with a more customer-oriented foundry model.

For now, the Atom-RosaicLabs arrangement is best understood as a reported, narrowly defined opening rather than a redesign of the x86 market. Its importance lies in what it signals: Intel may be more willing than before to consider monetising selected elements of its processor expertise, provided it can preserve control over its most valuable intellectual property and translate technical access into durable commercial relationships.

Sources