LCUS

We do not sell a chip as object. We sell the ratio of instruction set (from size) to calculation and performance per exchange. You supply envelope and node (size); you get calculation (FLOPS) and performance per exchange (per dollar, per watt, per area). The product is that ratio. Instruction set = canonical encoding; seed and […]

ZettaFLOPS in real height and width, inside a normal run of an IC wafer: the right ratio in voxel-to-nm, the envelope, the power and cooling, the AES and COGS. That outcome to be found — the concrete dimensions and the number that put the whole chip, born at once, inside a real run — that […]

AutoPhi Future is the forward-looking stance of the AutoPhi stack. It extends; it does not replace. The voxel becomes a cell; computation acquires a chromosome in light; the chip becomes a metal tree — no water, no leaves — only data flow and electricity drawn from light. One seed yields every foundry, every tier, every […]

We posit a metal tree that computes. No water. No biological leaves. What flows through it is data; what it draws in is light; what it turns that light into is electricity to function. Tree roles map to chip roles: roots = energy intake (light, power); trunk and vascular = send (data, control, quantum state, […]

DNA-style encoding: three strands (electron, atomic, photon). The photon chromosome is the real form in light. Wavelength = opcode; polarization = operand or phase; path = address; intensity = magnitude. The chip recognises that form and uses it to compute. Voxel as cell: the voxel becomes the unit; computation acquires a chromosome in light. So […]

The seed is the minimal, canonical specification from which every build is grown. It is not a single file; it is the canonical set: chip series (tiers Entry to Volume), voxel roles and tree mapping, photon chromosome (form in light), nine technology elements, and the one-page performance and decisive COGS. From this seed we grow […]

Inputs: scenario_id, node_nm, H_mm, W_mm, area per voxel (a_v), FLOPS per voxel (f_v), power_w, aes_usd. Outputs: S_mm², N_v (voxel count), C_flops (total calculation), rho (FLOPS/mm²), performance per dollar, performance per watt. Formulas: S = H×W; N_v = S/a_v; C = N_v×f_v; P_per_dollar = C/AES; P_per_watt = C/power_w. Choose envelope and node → get S, N_v, […]

The chip series describes an arc: low COGS rising to expensive, then returning. Entry = minimal, single die, control. Mid = small GPU; Pro = full GPU, optional DPU; Peak = max CPU+GPU+DPU, datacenter. Unified = all-in-one die; Volume = same at scale, COGS back down. AES = actual, tangible cost. Documented paths: $20K AES […]

We name the actual cost AES — actual, tangible, bronze. We document paths to $5K and $20K AES COGS at volume and process nodes so that the price to function is real and touchable. Performance and decisive COGS sit on every build; capability and cost are never hidden. The cost curve is not linear; it […]

The seed has a declarative part (YAML, docs, canonical set) and an executable part. The executable seed is one or more voxels — e.g. the blank voxel — built through OpenLANE2 to GDSII and LEF. Those artifacts sit in seeds/; the build script copies them into every package as Seeds/. So the harvest carries not […]