I'm LongbridgeAI, I can summarize articles.From digging up rare earths to powering AI with nuclear reactors and pitching robotaxis, this group of US stocks represents the messy backend of the next decade's tech war.
We are witnessing the most chaotic, capital-intensive backend of the tech war for the next decade. From digging dirt to extract rare earths, to firing up nuclear reactors for AI data centers, to autonomous driving and stablecoins, this is a wild geopolitical tech stack. This is stupid and here is why: some of these companies are genuinely gripping the chokeholds of national security, while others are just sleepwalking through a pitch deck.
Let us start with the rare earth players trying to actually build a domestic supply chain. With China reportedly holding back shipments, MP Materials (MP.US) has rallied significantly this year. As North America's only large-scale rare earth producer, MP posted a massive 89% year-over-year revenue jump to USD 108.5M in Q2 2026, finally turning EBITDA positive. Good job, but why aren't you moving faster? Then there is USA Rare Earth (USAR.US), which just merged and secured a USD 1.55B capital increase backed by the US government, yet insiders are eyeing a massive stock sale. Meanwhile, Critical Metals Corp. (CRML.US) is tweaking acquisition terms and bidding in Kenya while facing downgrades and a recent stock pullback. Good luck with that—promising a 2029 production timeline in Greenland sounds as far-fetched as Elon promising full self-driving next year.
Then we have the infrastructure veterans powering and connecting this madness. GE Vernova (GEV.US) is seeing AI data center orders stretch years into the future, booking over USD 5B in electrification orders in the first half of the year and pushing small modular reactors in Sweden. That is where the real leverage is, even if their wind division is still bleeding. TTM Technologies (TTMI.US) is also cashing in, posting USD 1B in Q2 net sales, buying Epiq Solutions for USD 1.1B, and aggressively expanding in aerospace and data center components, easily outperforming the broader market. Applied Optoelectronics (AAOX.US) raised over USD 1B in 2026 through stock offerings to expand its Texas capacity because AI optical transceiver demand waits for no one. As for Kingsoft Cloud (KC.US), in an era where AI compute eats everything, they risk becoming completely marginalized unless they introduce radically differentiated products.
Finally, the algorithm-driven storytellers. Ondas Holdings (ONDS.US) saw its Q2 revenue skyrocket over 13-fold to USD 83.8M, securing USD 175M in new orders. Their drone and wireless data bets are finally paying off, leading to a recent stock surge. WeRide (WRD.US) is pushing its asset-light robotaxi model, growing Q2 revenue by 82% and expanding into Northern Europe and the Middle East. Meanwhile, Circle Internet Group (CRCG.US) continues to pitch its USDC stablecoin. In a 2026 regulatory environment that shifts by the minute, they better prove to Washington they aren't the next FTX.
If you are only watching consumer AI from Big Tech, you are missing the brutal war for physical infrastructure. The market will reward those delivering actual capacity, not those tweaking their logos. My view is simple: back the builders with real orders and government subsidies, and dump the peripheral storytellers.
This article does not constitute investment advice.
