Humanoids can’t work alongside humans without a safety layer, and FORT Robotics has been building that layer since 2018. But its SPAC merger values $11.6M of 2025 revenue at 48x EV/sales.
Ambarella is a $2.82B edge AI chipmaker trading at 5.8x the FY2027 revenue guide, down from 8.4x in May. Two new distribution channels are sized at ~$500M each over seven years and X7 is its first standalone AI accelerator.
Agility will be the only listed U.S. humanoid pure-play, with a factory rated for 10,000 robots a year and paid deployments at Amazon, GXO, Schaeffler, Toyota, and Mercado Libre. However, at $13.40/share, you're paying $4.51B for just $1.78M of FY2025 revenue.
Fuel Tech trades at 1.6x sales with no debt and $29.6M of cash, ~70% of its market value. Strip the cash out and the operating business is worth ~$13.8M, with the data center pipeline as a free option.
A two-brand, share-gaining footwear business at ~12x earnings, ~40% ROE, and ~$1.1B of FCF, priced like it's done growing. Whether it's cheap or a value trap comes down to whether HOKA keeps its pricing power.
Reviva pushed its Phase 3 back a year, and the stock is down ~27% since my June write-up. Two developments matter more: (1) a patent filing that stretches exclusivity to 2046 and (2) evidence Teva has studied brilaroxazine since 2023.
Five quarterly updates on names from the write-up archive. I compare against guidance and discuss what actually matters underneath the headline numbers.
A US$28M company owns the largest known undeveloped indium resource in the US. The US imports 100% of its indium, China controls 70% of supply, and the Pentagon has started soliciting purchases for the stockpile.
A very busy earnings week across the names from past write-ups. The market mostly traded the headline while the filings told a more specific story.
A fresh look at five names from the archive, plus a quick recap of the market backdrop. For each name, I cover what happened to the business while the stock moved, where the mispricing has widened, and whether I'm adding at current prices.
SEALSQ trades at $2.43, just 11% above its net cash, so you pay close to nothing for an $18M secure-chip business and get its post-quantum and quantum-computing story on top.
Fluence cut guidance twice in one year and still missed. The stock now trades under 1x FY2026 revenue with a record $5.6B backlog, two hyperscaler MSAs, and a spot in NVIDIA's AI factory reference design.
Enovix makes the "highest energy density smartphone battery ever independently tested" and already ships cells for smart eyewear and drones. But smartphone qualification is two years behind, continued burn and dilution are inevitable, and insiders have only sold.
A US manufacturer selling into AI memory/logic chips, photonics, and defense batteries is going public at a $1.2B valuation. It's backed by a $100M DOE grant, Samsung SDI building its factory, and a 2028 law banning the US military from buying Chinese battery cells.
China choked off tungsten, and the miners already ran 10x. The hardest input to secure is high-purity powder, and chipmakers are now switching to molybdenum. One overlooked American name is positioned for both.
A record $783M quarter, a first-ever buyback, and open-market buys from the CEO, COO, and a director, all built on the Alani Nu and Rockstar acquisitions. But flagship CELSIUS slowed to 6% and gross margin slipped to 48.3%
The only U.S. company producing two commercial-purity heavy rare earths from ore, the kind almost no one outside China can make. Both are critical to defense and the AI buildout, and the U.S. government just committed up to $725M to back it.
Reviva trades below its cash and just dropped to the OTC market, but it owns a schizophrenia drug that topped Caplyta's Phase 3 numbers, the kind big pharma buys for billions
An Australian neocloud signed more than $2.2B of five-year, take-or-pay GPU contracts against $1.6M of 2025 revenue. A short seller calls them phantom. Oaktree and NVIDIA are funding the buildout anyway
5x earnings, a 14.6% distribution yield, and net cash worth a third of the market cap. But each one leans on a bird-flu spike that's reversing
QXO loses money, dilutes every quarter, and just levered up for a $17B acquisition. It might still be the best operator bet on the market
Q1 revenue came in a third short of plan, but the contracted backlog hit ~$1.45B and the ARR target rose to $400M. From here, it's all deployment
The restaurant operating system is more profitable than ever and pushing into enterprise, retail, and AI agents, but a consumer pullback, rising memory-chip costs, and tougher competition have halved its multiple
Ambarella owns the edge AI layer NVIDIA doesn't touch and just turned profitable. The new $800M Hanwha LTA and 15+ robotics design wins (including humanoids) are still ahead
PowerBank's history is ugly and it operates just 35 MW of solar today, but its sites, permitting relationships, and a recent LOI open a real path into the AI buildout's bottleneck: power delivery