From Washington: In a development that carries weight well beyond Silicon Valley, Intel has formally introduced its Xeon 6+ processors, codenamed "Clearwater Forest", to the world at MWC 2026. The chips pack up to 288 energy-efficient Darkmont cores and are the first data centre CPUs made on Intel's 18A fabrication process, a 1.8nm-class node the company has spent years and tens of billions of dollars bringing to production. For a company that has spent the better part of a decade playing catch-up to Taiwan's TSMC and rival AMD, the announcement carries the unmistakable weight of a corporate comeback bid.
The architecture itself is genuinely ambitious. The Xeon 6+ combines 12 compute chiplets containing 24 Darkmont cores per tile, produced using 18A manufacturing technology, two I/O tiles made on Intel's older Intel 7 node, and three active base tiles on the Intel 3 fabrication process. The compute tiles are stacked atop the base dies using Intel's Foveros Direct 3D technology, with lateral connections enabled by Intel's EMIB bridges. The result is not a monolithic slab of silicon but a carefully engineered mosaic of chiplets, a design philosophy that keeps manufacturing costs in check while pushing total core count to record-breaking territory.
Intel has aimed the Xeon 6+ primarily at telecom, cloud, and edge AI workloads, featuring Advanced Matrix Extensions (AMX), QuickAssist Technology (QAT), and Intel vRAN Boost technologies. The efficiency gains claimed are striking. In testing by Ericsson, a single Xeon 6990E+ Clearwater Forest chip with 288 cores delivered a 38% reduction in runtime rack power, more than 60% greater performance per watt, and 30% higher overall performance versus a dual-socket Xeon 6780E Sierra Forest platform carrying the same core count. For hyperscalers and telcos whose power bills stretch into the tens of millions of dollars annually, numbers like those are not marketing gloss; they represent genuine cost savings at scale.
The 18A node itself sits at the heart of Intel's wider strategic ambitions. The 18A process is designed to prove Intel can create a compelling CPU architecture and manufacture it internally on a node competitive with TSMC's best offerings; it is also the first 1.8nm-class process to enter high-volume production anywhere in the world, preceding TSMC's N2 by weeks or months. The node introduces Intel's second-generation RibbonFET gate-all-around transistors and the industry's first mass-production-ready backside power delivery network, PowerVia, together enabling up to 25% higher performance or 36% lower power consumption compared to Intel 3, alongside a roughly 30% increase in transistor density.
Those are impressive numbers on paper. The more uncomfortable reality, however, is that Intel's foundry business remains under heavy financial pressure. Intel's advanced 18A node is now in high-volume production at its new Arizona fab, but no major outside customers have yet emerged. Unlike TSMC, which only makes chips for outside clients, Intel also makes devices powered by its own chips, positioning it as a potential competitor to the very customers it hopes to attract: as one industry analyst put it, "if you're an Nvidia or AMD or Qualcomm or Broadcom, do you really want to put your secret sauce into a manufacturing operation where you're giving Intel access?"
There is also the question of yield. Efficient production of complex chips requires consistently high proportions of usable silicon from each wafer. While the official start of 18A production is a win for Intel, making it technically first with a 2nm-class node in production, it still faces a potent opponent in TSMC; the new node represents catching up rather than pulling ahead. TSMC holds over two-thirds of the broader foundry market and its loyal customer base includes Apple, Nvidia, and AMD, all of whom have deep infrastructure commitments to the Taiwanese manufacturer.
For Australian policymakers and businesses, the story has a direct read-through. A more competitive American semiconductor sector, manufacturing advanced chips on domestic soil, strengthens the strategic case for US technology partnerships underpinning arrangements like AUKUS. Australia's own technology sector, and its growing investment in AI infrastructure and data centre capacity, also has a stake in whether American chip manufacturers can deliver genuine alternatives to Taiwan-centred supply chains that carry well-documented geopolitical risk.
Intel plans to launch Xeon 6+ in the first half of 2026. That timeline gives the company a narrow window to demonstrate that 18A can deliver in volume, not just in carefully curated benchmark comparisons. The architecture is genuinely innovative, the efficiency claims are backed by third-party testing from credible partners, and the manufacturing is happening on American soil at Fab 52 in Chandler, Arizona. But the semiconductor industry is brutally unforgiving of overpromising, and Intel has a recent history of both. Whether Clearwater Forest marks a genuine turning point or another chapter in a long recovery story is a question the market, and its customers, will answer over the months ahead. Reasonable observers can hold both the genuine technical achievement and the remaining commercial uncertainty in mind at once; that tension, more than any benchmark figure, is what makes this announcement worth watching.