Could the AI era mean a reboot of more traditional Silicon Forest activity…?

Silicon Forest is an admittedly odd moniker. (Says the guy with an even more oddly named blog.) But it’s also fitting. And it’s a reminder that fabs and wafers and lithography and embedded systems and all of the rest of the deeply unglamorous physical technology business are foundational to this region. So when a16z dropped State of Markets II — written by general partner David George who leads their Growth team — and the second section announced that capital has rotated out of software and back into hardware… well, I took notice.

CAVEAT: Usual disclaimer applies. I am not a market analyst. I do not have a thesis. I read these things through my incredibly myopic lens of the Oregon startup community, as is my wont.

What the report actually says

tl;dr Hardware and infrastructure are where the money will be going.

In all events, hardware and infrastructure have emerged as the apple of the market’s eye, after years of trailing in software’s wake. Both public and private equity is funding innovation in compute, memory, power, robotics, manufacturing, and defense, with an intensity we haven’t seen in decades (if not longer). The point being: atoms are so back.

What’s more…? Compute. Memory. Power. Robotics. Manufacturing. All points of interest — if not complete strengths — for Oregon and Southwest Washington.

Let’s start with the general numbers. Tech contributed roughly 76% of the S&P 500’s total earnings growth in 2026, as of late August.

GPU rental rates were supposed to fall off a cliff as supply caught up. They didn’t. Even the A100 — the one everybody had written off — is pricing at or above where it started the year.

And then of course there’s the transfer of wealth. Hyperscaler free cash flow is rapidly becoming semiconductor free cash flow. And that’s heartening. Because those are the parts of the stack that Oregon truly excels at making.

The Intel factor

tl;dr Two of the most interesting compute companies in the country are being built by people who spent their careers here at Intel.

AheadComputing designs 64-bit RISC-V CPU cores aimed at AI and data center workloads. It was founded in Portland in 2024 by four former Intel CPU architects — Debbie Marr, who’s CEO and spent 33 years at Intel as a Fellow, along with Jonathan Pearce, Srikanth Srinivasan, and Mark Dechene. They raised a $21.5M round in February 2025 led by Eclipse, with Tenstorrent CEO Jim Keller among the investors. Then they raised another $30M this year, jointly led by Eclipse, Toyota Ventures, and Cambium, with Corner, Trousdale, EPIQ, MESH, and Stata also in. That’s $53M and roughly 120 people. In the Silicon Forest. Building processor cores.

Mueon is in Hillsboro, building something called Cubelets. They’re stackable modular units that pull compute, memory, power delivery, and thermal management into a single block. And importantly, a block that can be LEGO’d together with other blocks to right size compute. The ultimate goal…? Replacing racks of servers and all the cabling between them. They’re targeting gains of up to 10x on density, efficiency, and deployment speed. Founder and CEO Wilfred Gomes is an Intel Fellow who spent 27 years there working on 3D integration and Foveros packaging. The $15.5M round announced last September was led by Intel Capital, with the Geodesic Alliance Fund and Oregon Venture Fund participating, and Intel Capital’s Srini Ananth took a board seat.

You read that right. Compute, memory, power delivery. Three of the six words in that a16z passage. Inside one Hillsboro startup.

The power layer

tl;dr Compute has shifted from a chip problem to a power problem. And Oregon is working on power in ways that sound made up until you read the numbers.

Panthalassa is a Portland company putting data centers on floating nodes that convert wave motion into continuous onboard power. With the ocean handling the cooling. The pitch is availability: power up to 90% of the time, against 30% to 40% for offshore wind and about 25% for solar, at a target around two cents per kilowatt hour. Founded in February 2016. Tested in the wave tank at OSU. GeekWire reported in May that the company raised $140M led by Peter Thiel, with John Doerr, TIME Ventures, SciFi Ventures, Susquehanna Sustainable, Hanwha, Fortescue Ventures, Super Micro Computer, and Dylan Field among the co-investors. And Portland Seed Fund and the Intrepid Oregon Fund in there too. Which matters to me more than more widely recognizable names do.

Then there’s the volcano.

Newberry, south of Bend, has quietly become the most interesting power site in the state, and two companies are drilling into it. Neither one is from here. Mazama Energy raised $135 million in September, on the back of building the world’s hottest enhanced geothermal system out there last October — 629°F, at a pilot site near the volcano. Project Ceres is next, targeting 15 megawatts per well. The site itself, they now figure, could do 10 gigawatts.

Quaise Energy is at Newberry too, with $134 million of its own, vaporizing rock with millimeter waves from a gyrotron. Which sounds made up. It isn’t. Project Obsidian is 50 megawatts by 2030 and 250 after that, and they have already signed a hyperscaler for the first 50.

And then there’s the second volcano.

Endurance Energy just drilled a well into the seafloor at Axial Seamount, roughly 300 miles off the coast and about a mile down, going after hydrothermal fluid at 482°F. Their generator is named after a penguin. The readings come back over NSF research fiber already lying on the ocean floor between the seamount and the Oregon shore, which is part of why they picked the spot. Stanford’s geothermal director called the whole thing “a slightly dingbat idea, at least for grid generation.” He also said it has real potential for islands. Both of those are probably true.

So Oregon has two volcanoes being drilled for power right now. One in the high desert. One under the Pacific. Between Newberry and the seamount that’s $279 million and counting. And not one of the three companies is headquartered here.

That seems like an opportunity to me.

Across the river, PowerLattice in Vancouver is doing power delivery as chiplets for AI accelerators, claiming better than 50% power reduction. Peng Zou’s team raised a $25M Series A last November, jointly led by Playground Global and Celesta Capital, for $31M total — and the Playground side of that brings Pat Gelsinger, who is a general partner there now.

And that’s not even mentioning the solar, wind, and hydro power we have at our disposal. Power — and improving power production — could be the dark horse winner in this whole thing.

And still other promising pursuits

fJscaler is in Beaverton building mixed signal chips for optical networking inside AI data centers, chasing femtojoule per bit efficiency. Founded in 2022, about 55 people with roughly 30 of them here, with Jan Filip as CEO. Its lineage runs Tektronix to Maxim to Analog Devices to fJscaler, which is about as Silicon Forest a genealogy as you can draw.

And then there’s Agility in Salem, by way of OSU in Corvallis, which has been building a humanoid robot named Digit since before humanoid robots were a fundable category. Robotics is in the a16z list too. Agility has been at it for a decade.

And lest we forget, FAST. A consortium of nearly 100 Oregon partners — universities, companies, community organizations, a dozen counties — won an NSF Regional Innovation Engine award called FAST (Frontiers of Advanced Semiconductor Technology). With Oregon State University as the administrative home. It starts at $15 million for two years, can unlock $45 million more in years three through five, and tops out at up to $160 million over a decade — but only by hitting milestones the whole way.

So I don’t want to get overly optimistic. But I’d be remiss if I didn’t say that the pendulum swing appears to be benefitting the Silicon Forest in ways we haven’t seen since the bunny suits and the Intel Inside campaigns. Here’s hoping that upswing continues.

For more or to make your own assessment about the potential for the Silicon Forest, read a16z’s State of Markets II.