KEY TAKEAWAYS
- H1 2026 revenue: $78.4M, up 99.3% YoY. Q2 alone: $50.1M, up 138% YoY and 78% from Q1. Full-year guidance raised to at least $200M.
- Q2 operating loss narrowed to $2.2M from $41.1M in Q1 — but $25.4M of H1 stock-based comp was front-loaded into Q1, making Q2 look cleaner than the run rate suggests.
- 2027 target of $450M–$500M is contingent on Yotta Phase 1 ($100M annualized), Yotta Phase 2 batch 1 ($250M), and NeutraDC initial servers ($75-80M) — all dependent on GPU delivery schedules, site power readiness, and customer workload migration.
CEO Jayesh Chandan closed the call with this: "We are no longer explaining what Gorilla intends to become. We are beginning to demonstrate it." Fair enough. The H1 numbers are real. Revenue doubled. The operating loss collapsed 95% quarter-on-quarter. Cash is $179M. The question worth asking — and the one most coverage glossed over — is how much of the 2027 guidance is already contracted versus how much depends on things that haven't happened yet.
What they reported
H1 2026 revenue came in at $78.4 million, up 99.3% from the first half of 2025. Q2 on its own was $50.1 million — up 138% year-over-year and 78% sequentially from Q1. The company had originally guided for $33 million in Q2, upgraded to $44 million, then beat that by another $6 million. Q2 operating loss: $2.2 million, down from $41.1 million in Q1. Operating cash usage for H1: $4.3 million, down 65% from $12.5 million in H1 2025. Cash at end of June: $179.4 million, up 82% from the Q1 close, bolstered by a $107 million five-year convertible note placed in June.
Beat or miss
Beat, clearly. The company beat its own upgraded guidance by 14% and raised the full-year floor from $137 million to $200 million. But here's the thing about Q1 versus Q2 — $25.4 million of H1 stock-based compensation landed in Q1, with over 80% of the H1 charge absorbed there. The Q1 operating loss of $41.1 million included that burden. Q2's much cleaner $2.2 million loss reflects the same business with the SBC overhang gone. That's not manipulation — it's how SBC accounting works. But it means the "95% loss reduction" headline is partly a comparison to an artificially heavy Q1, not a pure operating improvement.
The 2027 target — what's in it and what isn't
This is the part that matters most and got the least analytical attention. CFO Bruce Bower laid out the 2027 guidance construction explicitly on the call. Yotta Phase 1 — testing complete, equipment deliveries underway, $100 million annualized incremental revenue — that's in guidance. Yotta Phase 2, first batch — manufacturing expected complete within 30 days, $250 million incremental revenue — that's in guidance. NeutraDC initial 300 servers, delivery schedule firmed up, roughly $75–80 million annualized — that's in guidance. The remaining 700 NeutraDC servers, NeutraDC Phase 2 (875 servers), and Yotta Phase 2 second batch are not included in guidance. Those three pieces, if they come in, would push revenue well above $500 million.
So yeah. The guidance is conservative by design. Bower said it directly: "We don't include it in guidance if timing isn't exactly firmed up." The $450–500 million target is based on contracted, scheduled deliveries. The upside sits in the uncontracted pipeline. For investors, the relevant question isn't whether the guidance is achievable — the CEO says the equipment for Yotta deliveries is literally on a plane right now. The question is whether the execution risk events Chandan named — hardware timing, site and power readiness, customer workload migration — materialize and delay the recognition timeline.
The power problem nobody wrote about
Chandan said it almost in passing: "electricity seems to be of a bit of a problem." Then in the Q&A, he returned to it: "electricity has an inconvenient habit of being very essential." For a GPU-as-a-Service business deploying infrastructure in India, Indonesia, and Thailand, power availability and reliability is not a footnote. It is the critical path. You can have the GPUs. You can have the data center space. You can have the customer contract. If the power drops aren't in and tested, none of it recognizes as revenue. The company has contingencies — Chandan said engineering teams are working 24/7 on site preparation — but power readiness across five simultaneous sites in developing market geographies is a real execution variable that the headline revenue numbers don't capture.
What the gross margin story actually looks like
Current gross margins are low because the revenue mix is weighted toward hardware and initial deployment — lower margin phases of the project lifecycle. Once the infrastructure is commissioned and workloads are running, the GPU-as-a-Service revenue kicks in at 75% gross margins on Yotta and NeutraDC projects. The 2027 gross margin expectation — 40%-plus at the consolidated level — reflects a revenue mix dominated by those high-margin compute services rather than the hardware delivery revenue that's driving H1 2026. That transition from hardware-heavy to service-heavy revenue is the structural story the stock is priced on. The risk is a timeline slip that keeps margins suppressed longer than the guidance assumes.
The one thing beginners should take away
GRRR is a pre-revenue-model-proven company reporting real revenue growth from real infrastructure contracts. That's meaningfully different from most small-cap AI stories. But the $450–500 million 2027 target is not a forecast — it's a construction from contracted deliveries that each carry their own execution dependencies. Hardware timing, power availability, and customer workload migration are the three variables between here and there. The CEO has been transparent about all three. The market is pricing the outcome where all three resolve favorably. That might happen. It might not happen on schedule. That gap between the target and the dependencies is what you're betting on when you buy this stock.