Intel is sinking $19 billion in the red and taking steps to stem the bleeding. Falcon Shores therefore has to go after earlier postponements. As for processors, we should not expect too much in 2025 either.
Over the past twelve months, Intel had to swallow a $19 billion loss, quarterly results show. Interim CEOs David Zinsner and Michelle Johnston Holthaus must now right the ship, after the company surprisingly tossed CEO Pat Gelsinger out the door early this year. Gelsinger tried to right the ship with a multi-year plan, but quick results failed to materialize, causing him to fall by the wayside. However, an alternative strategy does not appear to be in sight before 2025.
Away with impending GPU ambitions
The blunt axe is being brought out. After years of procrastination, Falcon Shores must go. Falcon Shores saw the light of day as an ambitious superchip on the level of Nvidia’s Grace Hopper or AMD’s Instinct MI300A. Intel thus wanted to combine its self-developed Xe-GPU architecture with proprietary CPU cores into an assembly that was to be ideally suited for AI-related HPC workloads in the data center.
Falcon Shores was then postponed and ambitions reversed: the chip would become an exclusive GPU (analogous to the AMD Instict MI300X) but would definitely appear in 2025. Now it appears that no Falcon Shores GPU will leave the lab. The GPU will remain an internal project and should serve as a springboard for a new effort: Jaguar Shores. That successor should then entice the market sometime in the future.
The postponement is painful for Intel, which has been talking about Falcon Shores for years and is now losing face again. Moreover, competitors AMD and Nvidia are diligently developing, and the accelerator and GPU ecosystem is thickening without a major role for Intel. Who will be waiting for a third alternative in a few years. Holthaus is counting on plenty of interest.
CPUs: waiting for 2026
CPUs are Intel’s core business, but even there the news is not too good. Late last year, the company proudly announced that the first Clearwater Forest-Xeon chips had successfully rolled off its own 18A production line. Intel 18A is a production mode that should be competitive with the most punishing TSMC can offer today. Smaller transistors make for more efficient and cooler chips. Combined with good design, Intel can basically make a fist against AMD and Epyc with Clearwater Forest.
Only: Clearwater Forest will not appear this year. The new deadline for Xeon chips is the first half of 2026. Holthaus is referring to Sierra Forest, which is coming out this year, for the decision. That is an Xeon CPU with up to 288 compute cores, although it is not traditional P cores. Intel crammed the CPU with fuel-efficient E cores and positioned the thing as ideal for cloud workloads. The demand for chips with economical E cores without multithreading now turns out not to be as big as expected after all. Holthaus speaks of a niche market. The launch of Cleatwater Forest so close to that, Intel now thinks is not such a good idea.
Nova Lake and TSMC
According to the company, the Intel 18A production line is doing well. PC CPUs under the Panther Lake name should launch by the end of this year, but would not roll off the 18A belt in large volume until early 2026. That does imply that we don’t expect significant jumps in Intel’s lineup before 2025. The manufacturer is already pre-emptively admitting that this year’s Panther Lake launch will not exactly coincide with volume production.
After Panther Lake then comes Nova Lake in 2026, which Intel somewhat surprisingly does not want to build entirely in its own factories. Intel previously grabbed TSMC as a manufacturing partner. This is notable anyway, given that Intel has many of its own factories and wants to offer their capacity externally.
Downward spiral, upward trajectory
The financial results and accompanying strategy give little cause for cheer. The numbers at Intel are turning red and interesting products are either postponed or discontinued. This keeps Intel a bit behind the times. After all, AMD (and Nvidia) are not sitting still and Qualcomm is also launching new interesting PC chips.
Thus Intel remains in a downward spiral. The once untouchable CPU builder cannot stay in the market today with chips that clearly outperform the competition. Often the opposite is true, and alternatives from AMD and Qualcomm are better. Even its long-awaited grand entry into the world of data center accelerators remains dead letter, with at most some launches on the fringes. That weighs on Intel’s results, which are turning red. In turn, the interim CEOs are responding by delaying products, reducing the likelihood that new chips will gain an edge over the competition.
Bad news, then, although the stock market is not panicking. We have seen for quite some time that the correlation between what a technology company actually delivers, and the evolution of its share price, is far off. So despite the gloomy outlook, Intel’s share price is rising slightly.