Every Language.
One IDE.

Developers can't see
energy cost — in any language.

Whether you write Python, Java, COBOL, or Rust — no traditional editor tells you the energy cost of your code. You optimize for speed. The planet pays in watts.

"240 billion lines of COBOL still run the world's banks. Nobody knows what they cost in energy. Nobody has a path to modernize them safely." The legacy blind spot

One IDE for all of them.

Full LSP support for every language. Energy metering across all of them. See the full catalog →

Write in any language.
Migrate when ready.

The Lift engine converts code from any supported language to Lux (web) or Joule (systems). One button. Formal verification on the output. You decide when to migrate — Arc never forces it.

-- Arc Lift: COBOL to Joule
01 WS-TOTAL-AMOUNT    PIC 9(7)V99.
01 WS-TAX-RATE        PIC 9V99.
COMPUTE WS-TOTAL-AMOUNT =
  WS-SUBTOTAL * (1 + WS-TAX-RATE)

        ↓ arc lift --to joule

fn compute_total(subtotal: Decimal, tax_rate: Decimal) -> Decimal {
    subtotal * (Decimal::ONE + tax_rate)
}

Same operation.
75.88x less energy.

Arc shows you the energy cost of every operation in every language. When you're ready to save 75x, the Lift button is one click away.

import numpy as np

def process_readings(data):
    # No idea what this costs in energy
    result = np.mean(data) * np.std(data)
    return result

use energy::{Budget, Receipt};

// Compiler knows: 12.4 uJ worst-case
#[energy_budget(15_uJ)]
fn process_readings(data: &[f32]) -> Receipt {
    let mean = data.iter().sum::<f32>() / data.len() as f32;
    let std = std_dev(data, mean);
    mean * std
}