Ramin Hasani
Frontier Insights
Core Frontier Thesis: Frontier AI’s future decouples from centralized clouds; native edge intelligence driven by non-Transformer architectures (LFM2) delivers brain-scale efficiency, zero hosting costs, and real-time privacy across automotive and industrial endpoints.
Strategic Moves: Liquid AI bypassed brute-force scaling, leveraging automated hardware-in-the-loop operator searches (AFMD) to deploy sub-1GB models directly onto client hardware, securing enterprise validation with Mercedes-Benz and Shopify.
Risks & Warnings: Severe hardware-model overcoupling could bottleneck cross-platform agility; bridging the capability gap to frontier cloud models remains perilous; and industry-wide production adoption hovers at a fragile ~11%, demanding radical workflow restructuring.
Key Views & Dialogues
Intelligence on the Edge: Liquid AI’s Ramin Hasani on the Search for Device-Native Foundation Models
Liquid AI is targeting edge inference across phones, cars, and factories, where privacy, latency, energy, and workload economics create a market beyond cloud AI. Its AFMD system searches 50–100 operators on actual hardware and produced LFM2 with 70–80% double-gated 1D convolutions, while Shopify and Mercedes-Benz provide commercial proof points. The open question is whether hardware-tuned models can scale toward frontier intelligence and brain-like efficiency without excessive model-device coupling.
View summary (8)
- Liquid AI’s commercial thesis is that edge hardware is an underused inference market, not merely a cheaper copy of cloud AI. Hasani sized smartphones alone at roughly $500 billion annually; Labenz framed phones plus laptops as about $1 trillion of compute shipped each year, while the introduction used roughly $800 billion. Energy limits, privacy, latency, and workload economics all support Liquid’s aim to “build an intelligence layer on top of the diverse formats of hardware” already in pockets, cars, factories, and other devices.
- Liquid’s central technical finding is that the best architecture depends on scale, specialization, and the hardware constraint. Attention remains the richest, least-structured mechanism and may justify its (n^2) cost at frontier scale, while smaller or narrower models benefit from recurrence, convolutions, gating, or domain-specific dynamics. “The larger the network becomes, the more unstructured you can make it”; conversely, constrained systems can trade generality for substantially better speed and memory efficiency.
- The scientific lineage began with biologically inspired differential equations that produced striking control systems from only tens of neurons. Liquid networks parallel-parked a small car with 12 neurons, drove with 19, and flew a drone with 30; later work extended the approach to jets and other predictive systems. Their advantage was input-dependent dynamics and out-of-distribution adaptability—not magic or continual learning: “There’s no free lunch,” and the trained parameters remain fixed.
- Liquid’s automated foundation-model design system turns architecture selection into a hardware-grounded search problem. AFMD evaluates roughly 50–100 operators and hybrid combinations using an evolution strategy, actual target processors, memory and latency constraints, and around 100 downstream benchmarks rather than perplexity alone. After searches spanning roughly 10 million to 72 billion parameters, the lesson was “You have to give it to the algorithms,” including when the algorithms discard Liquid’s founders’ own preferred mechanisms.
- LFM2’s winning CPU design is surprisingly simple: 70–80% double-gated 1D convolutional layers, plus a smaller allocation to attention. The gate makes computation input-dependent; the convolution supplies a cheap, unstructured operator, replacing much of attention’s memory and quadratic cost without the elaborate hand-tuned machinery found in many alternative architectures. Hasani’s blunt account of the search result: “All of this has to go away” when extra gates and human-selected features fail the full efficiency objective.
- Commercial proof points suggest Liquid has moved beyond an architecture experiment. Hasani reported more than 1 million weekly Hugging Face downloads, the No. 5 position among U.S. organizations behind Google, Meta, Microsoft, and NVIDIA, more than 50 model instantiations used in enterprises, and only about 1,000 in-house GPUs. Shopify uses Liquid models in production across commerce workloads, Mercedes-Benz signed a contract for a roughly 600-megabyte in-car audio and visual system, and Labenz found the 1-billion-parameter Apollo model fast enough on an iPhone for private document search and classification.
- Silicon vendors may need to own a tunable intelligence layer, not stop at chips and kernels. Liquid is working with companies including AMD and Qualcomm around processor road maps, while Hasani points to NVIDIA’s Nemotron effort as evidence that models optimized for hardware can improve the enterprise proposition. Labenz challenged whether this creates excessive model-hardware coupling; Hasani’s answer was essentially, “Why do you want to change the model?”—provided the default is fast, tunable, and does not block the wider open-source ecosystem.
- The immediate local-agent opportunity is orchestration, while true intelligence-per-watt miniaturization needs new learning paradigms. A proposed local LFM2 24B A2B system can route requests, filter PII, invoke specialized models, and escalate difficult work to the cloud, but Hasani says no current local model matches frontier quality without fine-tuning; forthcoming tooling could create production models for tens to low thousands of dollars. Longer term, current architectures will not approach the brain’s roughly 20-watt efficiency because “intelligence for me is an emergent property,” requiring objectives that induce multiple ways of learning—not next-token prediction alone.