Intel wants to rebuild trust with its foundry customers, CEO Lip-Bu Tan said during his keynote speech at the company’s Intel Foundry Direct Connect event in San Jose.

Reiterating his commitment to making Intel Foundry a success by strengthening its R&D efforts and manufacturing capabilities, Tan said that while he was excited about the possibilities, there was still a lot for Intel to improve upon, noting that past achievements do not guarantee future success.

The conference comes less than a week after the company published its Q1 2025 results, for which Intel Foundry posted a flat loss of $2.3bn for the quarter, and builds on comments made by Tan on the earnings call, where he told analysts: “I will not be satisfied until we regain the trust of our customers, putting the company on a sustainable path of gaining shares, growing revenue, and delivering consistent returns for our shareholders.“

While the message of rebuilding trust, staying humble, and better engaging with ecosystem partners was threaded through the speeches given by Tan; SVP and GM of Intel Foundry, Kevin O’Buckley; and EVP & GM of Intel Foundry technology and manufacturing, Naga Chandrasekaran, the executives were also intent on demonstrating the company is still a major player in the chip industry through a slew of product announcements.

Intel 18A is now in risk production, with high volume manufacturing of the node slated for later this year at Intel’s Oregon fab. Two new variants of 18A were also announced, Intel 18A-P and 18A-PT, which the company said would help serve a broader set of foundry customers.

Intel 18A-P has been designed to deliver enhanced performance and will add longevity to the node, as it will be rule-compatible with 18A. Meanwhile, 18AP-T is optimized for data center, Edge AI, and XPU applications and can be connected to the top die using Foveros Direct 3D with a hybrid bonding interconnect pitch less than 5 micrometers.

Additionally, unlike Nvidia and AMD, which have committed to yearly hardware releases, O’Buckley said Intel’s customers were in support of a two-year product cycle, with 18A-P and 18A-PT expected to be available in 2026 and 2028, respectively.

The company further announced that the first production of its 16nm tape-out is in the fab and has started to engage with lead customers for its 12nm nodes and derivatives, which are being built in collaboration with UMC.

Intel said it has also started to engage with lead customers for its successor to 18A, Intel 14A, and has already distributed an early version of its Intel 14A Process Design Kit (PDK). Slated for risk production in 2027, Intel says 14A will have a 1.3 percent density increase over 18A and will use Turbo cell technology to provide a 15-20 percent performance per watt improvement, when compared to its predecessor.

Updates to its advanced packaging portfolio were also announced at the conference, including new system-level integration of 14A on 18A-PT using the company’s 3D stacking technology dubbed Foveros Direct, and embedded multi-die interconnect bridging, also known as 2.5D bridging.

The company further unveiled the addition of an EMIB-T 2.5D offering to support future high-bandwidth memory needs, alongside two new additions to its Foveros architecture: Foveros-R and Foveros-B, which the company says will provide additional efficient and flexible options for customers.

Finally, Intel launched two new programs within Intel Foundry’s Accelerator Alliance, Intel Foundry Chiplet Alliance and Value Chain Alliance, and announced a partnership with Amkor, set to be fully enabled by the end of 2026.

Given President Donald Trump’s desire to bring manufacturing back to the US, Intel was also unsurprisingly keen to hammer home its position as a US-based chip manufacturer, with Tan saying his firm was committed to working with the Trump administration to achieve a shared manufacturing goal of producing advanced semiconductors in North America. Additionally, the company announced that Intel 18A and Intel 14A research, development, and wafer production will all be US-based.

However, despite O’Buckley using part of his keynote speech to discuss how important a customer the US government is to Intel, any mention of the CHIPS and Science Act and the $7.68 billion the company was awarded under the program was conveniently missing.

And while, despite ongoing delays, Intel was keen to talk about increasing capacity at its manufacturing sites across its Arizona, Ohio, and Oregon fabs and advanced packaging facility in New Mexico, there was no update on the facilities in Germany and Poland, beyond a color-coded map noting they are still currently paused.

Tan might have only been at the helm for five weeks, but it already seems clear that Intel is under new management, as illustrated by the fact that the company has opted to update its famous “copy exactly” approach to fab building with what Chandrasekaran described as “continuous improvement.”

“The most important thing is we have to earn our customers' trust and we have to change our culture to have a customer-first mindset,” Chandrasekaran said.

“The change from a copy exact to continuous improvement mindset; the change from telling others to listening to our customers and suppliers; the change from measuring our success by what Intel delivers to measuring our success by what the customers see as Intel delivering for them. These are the changes that we are making to ensure we have a customer-first mindset in what we deliver.”