Cloud Selection

The cloud selection is the AWS vs the Azure vs the GCP for the AI, and the exit-cost question (L366) — the clouds (the AWS L261, the Azure L366, the GCP L366), the AI axes (the Bedrock L278, the Azure OpenAI L366, the Vertex L366, L366), the economics (the pricing L366, the commitment L285, the egress L285, L366), and the exit (the portability L377: the containers L288, the K8s L306, the OTel L346, L366) — the enterprise's (L380) home, chosen (L366).

The one-sentence interview answer: "The cloud selection is the platform's choice (L366). The clouds (L366): the AWS (L261), the Azure (L366), the GCP (L366) — each with the AI (L366): the AWS's Bedrock (L278), the Azure's OpenAI (L366), the GCP's Vertex (L366). The axes (L366): the AI services (L366) — the model access (L278) and the AI stack (L283); the economics (L366) — the pricing (L366), the commitment (L285) — the reserved (L285) — and the egress (L285) — the data out (L261); and the enterprise (L366) — the compliance (L371), the support (L366), the existing (L366) investments (L366). The exit (L377): the portability (L377) — the containers (L288), the K8s (L306), the OTel (L346) — the exit cost (L366) minimized (L377). The decision (L361): the scored table (L362) — the axes (L366) weighted (L362) — the ADR (L361). The AI shape (L173): the enterprise (L380) — the cloud (L366) hosting the L287 stack (L287) — the AI services (L366), the economics (L366), and the exit (L377), evaluated (L366)."

Think of the cloud selection as the city's location decision. The company (the enterprise, L380) chooses the city (the cloud, L366): the AWS (L261), the Azure (L366), the GCP (L366). The city's offerings (the axes, L366): the power plants (the AI services, L366) — the Bedrock (L278), the Azure's OpenAI (L366), the Vertex (L366); the taxes (the economics, L366) — the rates (L366), the commitments (L285), the tolls (the egress, L285); and the moving costs (the exit, L377) — the portability (L377): the standard containers (L288), the standard rails (L346). The company (L366) visits (the PoC, L366), costs (L368), and signs the lease (the ADR, L361). The city works because the plants are strong, the taxes are known, and the moving is cheap (L366).

   the city (the cloud, L366)
   ┌────────────────────────────────────────────────────────┐
   │ the cities (L366) — the AWS (L261), the Azure, the GCP │
   │ the plants (the AI services, L366) — the Bedrock       │
   │ (L278), the Vertex (L366)                              │
   │ the taxes (the economics, L366) · the moving (the      │
   │ exit, L377)                                            │
   └────────────────────────────────────────────────────────┘

The mental model is the city: the cities, the plants, and the moving (L366).

   the requirement (L359)
        │  the AI stack (L287) to host
        ▼
   ┌────────────────────── THE SHORTLIST (L366) ────────────────────────┐
   │  the AWS (L261) · the Azure (L366) · the GCP (L366)               │
   └──────────────────────────┬───────────────────────────────────────┘
                              ▼
   ┌────────────────────── THE AXES (L366) ─────────────────────────────┐
   │  the AI services (L366): the Bedrock (L278), the Azure OpenAI,    │
   │  the Vertex — the fit (L362)                                      │
   │  the economics (L366): the pricing (L366), the commitment (L285), │
   │  the egress (L285)                                                │
   │  the exit (L377): the portability (L377) — the containers (L288)  │
   └──────────────────────────┬───────────────────────────────────────┘
                              ▼
   ┌────────────────────── THE DECISION (L361) ─────────────────────────┐
   │  the scored table (L362) → the ADR (L361): the cloud (L366)       │
   └──────────────────────────────────────────────────────────────────┘

The flow is the decision: shortlist → axes → decision (L366).

• The clouds (L366). The AWS (L261), the Azure (L366), the GCP (L366) — each with the AI (L366).
• The AI axes (L366). The Bedrock (L278), the Azure OpenAI (L366), the Vertex (L366) — the model access (L148) and the AI stack (L283).
• The economics (L366). The pricing (L366), the commitment (L285), the egress (L285).
• The exit (L377). The portability (L377): the containers (L288), the K8s (L306), the OTel (L346).

• An enterprise engagement (L379). The cloud (L366) — the AWS (L261) vs the Azure (L366) — the scored table (L362).
• A model access decision (L148). The Bedrock (L278) vs the Azure OpenAI (L366) — the AI axis (L366).
• A cost review (L368). The commitment (L285) and the egress (L285) — the economics (L366).
• A migration (L377). The exit (L366) — the portability (L377) — the containers (L288).
• Anything enterprise (L380). The cloud's choice (L366) — the axes and the ADR (L366).

The through-line: the choice is the cloud's — the AI, the economics, and the exit (L366).

Say it in four moves:

1. The clouds. "The AWS (L261), the Azure (L366), the GCP (L366)."
2. The AI axes. "The Bedrock (L278), the Azure OpenAI, the Vertex (L366)."
3. The economics. "The pricing, the commitment (L285), the egress (L285)."
4. The exit. "The portability (L377) — the containers (L288), the OTel (L346)."

• The AI services are the fit (L366). The Bedrock (L278) and the Vertex (L366) — the model access (L148) — the L283 stack (L283) — the fit (L362) axis (L366).
• The commitment is the cost's (L285). The reserved (L285) — the steady load (L358) at the discount (L285) — the L285 lever (L285), cloud-shaped (L366).
• The egress is the hidden cost (L285). The data out (L261) — the L285 line (L285) — the multi-cloud (L377) traffic (L366).
• The portability is the exit's (L377). The containers (L288) and the OTel (L346) — the standard (L346) — the exit (L366) cheap (L377).
• The ADR is the record (L361). The scored table (L362) — the choice (L361) — the reviewable (L361) decision (L366).

• The familiar picked (L366). The known cloud (L366) without the axes (L366) — the AI services (L366) un-compared (L366).
• The AI axis ignored (L366). The infra (L366) compared (L366), the model access (L278) forgotten (L366) — the AI stack (L283) is the fit (L362).
• The egress hidden (L285). The compute (L366) compared (L366), the egress (L285) forgotten (L366) — the L285 line (L285) (L366).
• The lock-in ignored (L377). The proprietary (L366) services (L366) — the exit (L377) — the portability (L377) (L366).
• The ADR-less (L361). The choice (L366) un-recorded (L361) — the review (L361) impossible (L366).

• Compare the AI services (L366) — the Bedrock (L278), the Vertex (L366).
• Cost the economics (L368) — the pricing (L366), the commitment (L285), the egress (L285).
• Plan the exit (L377) — the portability (L377): the containers (L288), the OTel (L346).
• Weight the axes (L362) — the scored table (L362).
• Record the ADR (L361) — the choice (L366), reviewable (L361).

Q: Walk me through the cloud selection.

A: The platform's choice (L366). The clouds — the AWS (L261), the Azure (L366), the GCP (L366). The AI axes — the Bedrock (L278), the Azure OpenAI, the Vertex (L366). The economics — the pricing, the commitment (L285), the egress (L285). And the exit — the portability (L377).

Q: How do you compare the AI services?

A: The fit (L362): the model access (L148) — the Bedrock (L278) vs the Azure OpenAI (L366) vs the Vertex (L366) — the model's (L148) choice (L148) and the routing (L155); and the AI stack (L283) — the serverless (L283), the RAG (L280), the observability (L346). The AI services (L366) are the cloud's (L366) AI fit (L362).

Q: What's the exit cost?

A: The migration (L366): the proprietary (L366) services (L366) — the Bedrock (L278), the DynamoDB (L366) — vs the portable (L377): the containers (L288), the K8s (L306), the OTel (L346). The portability (L377) — the standard (L346) — the exit (L366) cheap (L377). The exit question (L366) is the long-term (L366) cost (L366).

Q: How do you decide?

A: The axes (L366) weighted (L362): the AI services (L366), the economics (L366), the exit (L377) — the scored table (L362) — the ADR (L361) — the choice (L366) reviewable (L361). The PoC (L366) — the golden set (L342) on each (L366) — the evidence (L364).

• What are the clouds (L366)?
• How do you compare the AI services (L366)?
• What's the exit cost (L377)?
• How do you decide (L366)?
• What's the commitment (L285)?

The senior read: the AI fit and the enterprise's existing stack (L366).

// The cloud selection (L366) — the axes and the ADR (L366).
// 1 · THE SHORTLIST (L366) — the clouds (L366).
const shortlist = [
  { id: 'aws',   ai: 'bedrock',  egress: '$0.09/GB', commitment: '1yr' },   // L261, L278
  { id: 'azure', ai: 'openai',   egress: '$0.087/GB', commitment: '1yr' },  // L366
  { id: 'gcp',   ai: 'vertex',   egress: '$0.085/GB', commitment: '1yr' },  // L366
];

// 2 · THE AXES (L366) — the weighted scores (L362).
const axes = {
  aiServices:  { weight: 0.4, score: (c) => aiFitScore(c.ai) },    // L366, L362
  economics:   { weight: 0.3, score: (c) => costScore(c) },        // L368
  exit:        { weight: 0.3, score: (c) => portabilityScore(c) }, // L377
};

// 3 · THE SCORE (L362) — the weighted sum (L362).
const ranked = shortlist
  .map((c) => ({ ...c, total: score(c, axes) }))
  .sort((a, b) => b.total - a.total);
// the aws wins — the Bedrock (L278) and the L287 stack (L287)

// 4 · THE ADR (L361) — the choice recorded (L361).
//   ADR-032: the AWS — the Bedrock (L278) fit, the commitment (L285),
//   the exit (L377) via the containers (L288) and the OTel (L346)

What the reader must SEE — the choice, applied:

  the 3-cloud shortlist       → the candidates (L366)
  aiServices 0.4 + economics + exit → the axes (L362)
  the weighted total          → the score (L362)
  aws wins → the ADR          → the record (L361)

  The AI, the economics, and the exit — evaluated (L366).

• The AI fit is the stack's (L366). The Bedrock (L278) — the L283 stack (L283) — the fit (L362) axis (L366).
• The commitment is the cost's (L285). The reserved (L285) — the steady (L358) at the discount (L285).
• The egress is the traffic's (L285). The data out (L261) — the multi-cloud (L377) — the L285 line (L285) (L366).
• The portability is the exit's (L377). The containers (L288) — the migration (L366) cheap (L377).

• The cloud selection = the cloud's choice (L366): the clouds, the AI axes, the economics, the exit.
• The clouds: the AWS (L261), the Azure (L366), the GCP (L366).
• The AI axes: the Bedrock (L278), the Azure OpenAI, the Vertex (L366).
• The economics: the pricing (L366), the commitment (L285), the egress (L285).
• The exit: the portability (L377) — the containers (L288), the OTel (L346).

CLOUD SELECTION = the AWS vs the Azure vs the GCP for AI

THE CLOUDS (L366)
  the AWS (L261) · the Azure (L366) · the GCP (L366)
  each with the AI (L366)

THE AI AXES (L366)
  the Bedrock (L278) — the many models (L278)
  the Azure OpenAI (L366) — the OpenAI (L152)
  the Vertex (L366) — the Gemini (L154)
  the fit (L362) — the AI stack (L283)

THE ECONOMICS (L366)
  the pricing (L366) · the commitment (L285) — the reserved (L285)
  the egress (L285) — the data out (L261)

THE EXIT (L377)
  the portability (L377): the containers (L288), the K8s (L306),
  the OTel (L346)
  the exit cost (L366) — minimized (L377)

THE DECISION (L361)
  the scored table (L362) — the weighted axes (L362)
  the ADR (L361) — the choice (L366)

INTERVIEW, 4 MOVES
  1 clouds  "the AWS, the Azure, the GCP (L366)"
  2 AI axes "the Bedrock, the Azure OpenAI, the Vertex (L366)"
  3 economics "the pricing, the commitment, the egress (L366)"
  4 exit    "the portability (L377)"

You now hold the choice: the clouds, the AI axes, the economics, and the exit — with the plants compared and the moving cheap. The company signed the lease — and the city's address is in the logbook (L366).

Next: the named trade-off is the senior deliverable — Architecture Trade-offs (L367).