The $60 billion resource hiding in space, and the start trying to mine it | E2268

Summary

• Astroforge is developing low-cost, replicable spacecraft to mine platinum group metals (PGMs) from near-Earth asteroids, targeting a $60B annual market where these materials are geographically concentrated and scarce on Earth.
• The company's economic thesis: a mission (Deep Space 2) costs ~$10.4M to launch, with the potential to return up to 1,000 kg of material worth ~$105M at current prices, targeting 90% gross margins vs. ~14% for the best terrestrial mines.
• Key technical approach: targeting small (~200m) magnetic, metal-rich asteroids; using directed energy (lasers) to remove material and magnetism to sort it; landing via magnetic docking; operating on ultra-low data rates (~400 bps) at the asteroid to reduce cost.
• The company is in a high-risk, iterative development phase: its first deep-space mission (Odin) failed due to a solar panel deployment issue; its next mission (Deep Space 2/Vestri) is scheduled for November 2025 on a Falcon 9.
• Templar operates a subnet on BitTensor, using a decentralized, permissionless network of GPUs to train large AI models via a competitive incentive mechanism that rewards miners for reducing loss on training runs.
• Templar's value proposition: it trained a 72B parameter model for ~$2-3M distributed to miners, offering a potential path to "sovereign AI" for entities that cannot afford centralized, data-center-scale training costs.
• A key innovation in development is "heterogeneous sparse local training," which aims to allow training across any GPU type and network topology, potentially utilizing a vast, underutilized global inventory of non-latest-gen GPUs.
• OpenOats is an open-source, local-first notetaking agent that provides real-time meeting insights, fact-checking, and suggestions (e.g., from YC's Startup Manual) with a focus on privacy by avoiding sending transcripts to third-party LLMs.
• A meta-observation on media: picking an extreme ideological niche (e.g., far-left or far-right) can be an effective audience capture strategy, as seen in debates around publications like Wired and Jacobin.
• An observation on IP economics: Warner Bros.' planned $5.6B Harry Potter TV series remake is justified by the lifetime value of capturing super-fans globally via HBO Max/Paramount+ and driving downstream theme park and merchandise revenue.