How Promise.all Works in a Single-Threaded JavaScript World

If JavaScript is single-threaded, how can Promise.all() handle multiple requests at the same time?

The short answer:

JavaScript is single-threaded for executing code, but I/O is handled outside the JS thread.

---

🧠 What “single-threaded” really means

• Only one piece of JavaScript runs at a time.
• JS does NOT block while waiting for network, timers, or disk.

---

⚙️ Step-by-step: What happens with Promise.all

await Promise.all([
  fetch("/a"),
  fetch("/b"),
  fetch("/c"),
]);

1️⃣ Requests start immediately

const p1 = fetch("/a");
const p2 = fetch("/b");
const p3 = fetch("/c");

• Promises are created instantly.
• Network work starts in the background.
• JS thread is free.

---

2️⃣ Network runs outside JS

• Browser or Node runtime handles requests.
• Requests run concurrently.
• JS thread does nothing during this time.

---

3️⃣ Promise resolution uses microtasks

• When a request finishes, its Promise resolves.
• .then() callbacks go to the microtask queue.
• JS executes them one by one.

---

4️⃣ Promise.all resolves last

• Waits for all Promises to resolve.
• Rejects early if one fails.

---

🔄 Timeline

JS Thread: start A → start B → start C → (free) → run handlers
  ├─ Network: A request → A done
  ├─ Network: B request → B done
  └─ Network: C request → C done

---

🧠 Mental Model

• JS thread = single chef
• Network = multiple delivery drivers
• Promise.all = wait for all deliveries

---

⚠️ Important note

Promise.all() does NOT make CPU work parallel.

Promise.all([
  heavyTask(),
  heavyTask(),
]);

This still blocks the UI.

---

🧱 Browser vs Node

Layer	Responsibility
JS Thread	Executes callbacks
Browser APIs / libuv	Async I/O
Promises	Coordination

---

✅ Summary

Concept	Meaning
Single-threaded	One JS execution at a time
Concurrent	Multiple async ops in flight
Promise.all	Waits for all async results
Parallel JS	❌
Parallel I/O	✅

---

✨ Final takeaway

JS coordinates work — it doesn’t do it all itself.

Promise.all works because the heavy work happens outside the JS thread.

---

Originally published on Bitlyst.