Most DePIN tokens are a story still looking for revenue. $GRASS has the reverse problem → a business already selling to frontier labs, and a token that isn’t yet wired to most of what it earns 1) The edge > 8.5M+ users’ home connections across 190+ countries reach sites that block datacenter proxies > Data sold as training sets to frontier labs, Mag7 names and private AI companies 2) The business > $17M revenue in 1H26, roughly 6x year-on-year > FY26 guided at $65–75M against $2–3M of monthly expenses 3) Contents API & LCR model > Turns any public URL into model-ready text for agents, which moves GRASS from one-off dataset sales towards recurring, per-request demand > Grass’s own test: 97% retrieval vs 84% Firecrawl, 76% Parallel, 66% Exa. >4M+ requests a day shortly after launch. 4) The valuation > @FalconXGlobal base case: ~$165M revenue in 2027, ~$355M in 2028, with live retrieval near 40% of it > $GRASS trades at ~10x FDV to FY26 guidance, below historical private AI data peer multiples 5) The bear case > Most revenue gets reinvested - buybacks so far are small and discretionary > USDC node payouts kill emissions sell pressure but give the token no/arbitrary role in payments > The benchmark is self-run on 100 sites and unverified > Foundation-held supply makes the real FDV higher than the headline The play underneath it is that @grass routes a formal share of cash flow to the token while it still trades at a discount to its peers.