As crazy as this might sound, the global semiconductor shortage and the U.S.-China trade war couldn’t have come at a better time for Intel.

The shortage is expected to last at least through the end of 2022, and for the chipmaker-turned-foundry-operator, this is a very good thing. Intel needs time to bring new foundries online, retrofit existing ones, and, critically, squash any lingering bugs in its 7-nanometer manufacturing process before it goes online in, you guessed it, 2023.

Growing national security concerns will likely play to Intel’s benefit as well.

Intel is one of only a handful of U.S. chipmakers producing chips domestically. In fact, most chipmakers, U.S. or otherwise, are fabless and contract with companies like Taiwan Semiconductor Manufacturing Co. (TSMC), Samsung Electronics, or Global Foundries for manufacturing. Because of this, it would surprise me if Intel wasn’t on the shortlist for government subsidies and contracts, which reportedly could top $52 billion.

And while lawmakers claim these subsidies will help bring a speedy end to the chip shortage, let’s be honest. All the money in the world isn’t going to change the fact that building new fabs takes time.

Intel’s $20 billion Arizona chip fabs aren't scheduled to come online until 2024 at the earliest, and I suspect it could take even longer before they reach full capacity.

That’s not to say these subsidies aren’t warranted. Foundries not only take years to build, they are exceptionally expensive. If the government is serious about shoring up U.S. semiconductor supply, fronting the cash on an inherently low-margin, low-yield business isn’t a bad place to start.

And that’s really what this bill is about. It’s about ensuring a ready supply of domestically designed and built semiconductors for military and government applications.

Of course, Intel’s reinvention under the leadership of newly appointed CEO Pat Gelsinger isn’t happening in a vacuum. The company’s long-time rival AMD hasn’t been sitting still, and in recent years has arguably surpassed Intel technologically.

With the launch of EPYC, AMD moved to a chiplet architecture that essentially glued together several smaller CPU dies on a single package to achieve higher core counts and better yields. AMD’s EPYC 3 processors, announced this spring, offer up to 64 cores and 128 threads compared to Intel’s new Xeon Scalable chips, which now top out at 40 cores and 80 threads. That’s a pretty significant core deficit.

Meanwhile, AMD’s partnership with TSMC has allowed the company to bring 7-nanometer chips to market a full five years ahead of Intel’s projected launch date. Remember, Intel’s most advanced chips that it launched this spring are built on a 10-nanometer process. It’s 7-nanometer silicon won’t even hit the market for another two years.

There is no other way to spin this: Intel is falling behind.

Even if you buy into the argument that Intel’s 10-nanometer process offers comparable performance to TSMC’s 7 nanometer, you’re still comparing a brand new process to a 3-year-old one.

A much more revealing comparison would be between Intel’s 10-nanometer and TSMC’s 5-nanometer processes. We’ll actually get to see that comparison sometime next year when AMD launches its next-generation Zen 4 microarchitecture, which will be built on TSMC’s 5-nanometer process.

Here again, the semiconductor shortage, or more specifically, extremely high demand for TSMC’s foundry capacity, could benefit Intel in the meantime. Since Intel doesn't have to fight for foundry capacity — It owns its own fabs after all — it may help the chipmaker keep up with consumer demand where AMD/TSMC may struggle.

It should be noted that TSMC has committed $100 billion to expand its foundry operations, including a $12 billion foundry operation in Arizona. However, as I said before, it’s going to be years before the facility comes online.

Setting aside the chip shortage and stiff competition from rival chipmakers, there is still a lot we don’t know about Intel’s trajectory.

Gelsinger has been at Intel for less than six months and has already taken dramatic moves to reshape the company’s future. His decision to open its foundry operations is arguably the most profound.

With that said, it will likely be several years before we start to see Gelsinger’s influence on the company’s products. Semiconductors have long development cycles, and the next few generations of chips almost certainly entered development long before Gelsinger took over.

Gelsinger has also signaled an increased reliance on TSMC for high-end chip manufacturing, but when the first Intel CPUs manufactured by TSMC will hit the market remains to be seen.

With that said, the former VMware CEO has offered a glimpse of what is to come, and it bears a striking resemblance to AMD’s EPYC.

What we know is Intel is investing heavily in bringing its Foveros packaging process to market. Earlier this year, the company announced a $3.5 billion project to retrofit its New Mexico plant to produce chips using the technology.

Foveros is somewhat similar to AMD’s compute complex chiplet design in that it breaks down the CPU into modular components. But instead of packaging chiplets — or tiles as Intel prefers to call them — side by side, Foveros enables Intel to stack them vertically.

This packaging technology could enable Intel to package a combination of internally developed and third-party intellectual property on a single chip.

Not only should this technology allow Intel to better compete with AMD, but it will also help the chipmaker meet growing demand for specialized processor architectures — a market that's currently dominated by Arm chipmakers like Marvell and hyperscale cloud providers like Amazon. However, AMD isn't far behind. The rival chipmaker announced a similar technology coming to its own CPUs.

The future of Intel’s process designs is an even bigger enigma at this point. What comes after the company’s long-delayed 7-nanometer process? We know Intel is working with IBM on next-generation manufacturing technologies. We also know IBM has developed a 2-nanometer process node that can fit 50 billion transistors into a package the size of a fingernail.

Does this mean we’ll see Intel make the jump from 7 nanometers to 2 nanometers? The timing looks promising with IBM’s new process coming online in 2024. That puts it right in line with Intel’s new fabs.

So while Intel has a long and difficult road ahead, I’m optimistic that the best Intel has to offer is yet to come.

Editors note: This story has been updated to correct the anticipated launch date of AMD's Zen 4. Zen 4 is slated for release sometime in 2022. We regret the error.