Semiconductors

Intel looks to the future with Turbo Cell Technology for chips

The Californian company presented Intel 14A. And meanwhile, O'Buckley unveils the goal: 'To become an AI service company'.

from our correspondent, Biagio Simonetta

Un momento del keynote

2' min read

2' min read

SAN JOSE (CALIFORNIA) - After raising the bar with the 18A node, Intel is already looking further ahead and unveiling the first details of 14A, its next-generation semiconductor manufacturing process. The announcement came during Intel's Foundry event in San Jose, where the Californian giant outlined how 14A will bring a further leap forward in terms of density, energy efficiency and performance, thanks in part to a new and still little-known technology called 'Turbo Cell'.

According to reports, the new 14A process will deliver a 15%-20% performance-per-watt increase over the already advanced 18A node, taking advantage of both architectural and layout improvements. Like its predecessor, the 14A will adopt second-generation RibbonFET transistors (gate-all-around) and backside power delivery, key technologies to increase performance and reduce power consumption.

Loading...

But the real innovation is Turbo Cells, a solution designed to further improve CPU and GPU operating speeds.

It must be said that the company has not yet provided many details about this technology, but explained that Turbo Cell allows designers to combine high-performance cells with low-power cells within a functional block, creating a more efficient distribution.

This flexible architecture is expected to optimise CPU maximum frequencies and GPU critical paths, better adapting to the specific needs of each chip. According to engineers, the key will be the ability to balance performance and consumption in an even more granular way than previous nodes. And this, especially in terms of artificial intelligence, could be a remarkable step.

Node 14A is planned for 2027, and will be joined by a variant called 14A-E, which will introduce further functional extensions for advanced applications.

Meanwhile, the company led by Lip-Bu Tan - who opened the keynote at the Californian event - is already distributing the Process Design Kit (PDK) to customers, i.e. the design files needed to develop customised chips with the new technology. And he confirmed that several customers have expressed interest in test chips on 14A.

Meanwhile, the focus remains on 18A, which will enter mass production in the second half of this year with 'Panther Lake' chips for the notebook market, and in 2026 with Clearwater Forest processors for data centres.

A further development, the 18A-P node, expected in 2026, designed for a wider range of foundry customers, even beyond classic CPU computing, was also announced.

During the event, Kevin O'Buckley, SVP of Intel Foundry Services, emphasised that the company's long-term strategic goal is to become an 'AI services company' by aligning with the growing global demand for artificial intelligence chips. And a three-dimensional image of an enterprise-class GPU, similar to those developed by Nvidia, was also shown, reinforcing Intel's intention to compete on the AI hardware front.

Copyright reserved ©

Brand connect

Loading...

Newsletter

Notizie e approfondimenti sugli avvenimenti politici, economici e finanziari.

Iscriviti