Consulting

01 Discover

AI Infrastructure Planning & Architecture

What should we run, where should we run it and what should we buy, if anything?

  1. Discover Planning & Architecture Know what you need before you buy.
  2. Validate GPU Workload Assessment Prove what your hardware can deliver.
  3. Optimise Performance & Efficiency Get more from the infrastructure you own.
01 Discover. Three stages, each independently purchasable. Start where you are, or scope two or more together as one programme.

Who it is for

Buyers planning private AI deployments, replacing or expanding GPU capacity, assessing edge inference or questioning whether new hardware is necessary.

What's included

  • Discovery of use cases, model requirements, workloads, concurrency, latency, data handling and operating constraints.
  • Review of existing infrastructure and reuse opportunities.
  • Hardware-agnostic evaluation of suitable options, including NVIDIA, AMD and Apple Silicon where appropriate.
  • Preliminary capacity, memory, power and operational sizing.
  • Indicative total cost of ownership, risks and procurement roadmap.
  • Written architecture recommendation and walkthrough.

What's not included

  • Purchasing or reselling hardware.
  • Physical installation.
  • Production benchmark campaigns.
  • Performance commitments for equipment that has not been tested.
  • Formal regulatory certification.
  • Extensive comparative benchmarking, which needs a separately priced scope.

Price and timing

From $10,000 AUD ex GST

Typically 5-6 days of engineering effort over 2-3 weeks.

The starting fee assumes a bounded workload, site and platform scope. Additional complexity, architectures, workloads, sites or testing are quoted separately. Elapsed time is not engineering effort.

Discuss your workload

Frequently Asked Questions

Do we have to buy new hardware?

No. Recommendations are hardware-vendor independent and may be to reuse existing equipment, or not to buy new hardware at all.

Which hardware platforms do you assess?

NVIDIA, AMD, Apple Silicon and other appropriate architectures, assessed on their merits against your workloads, cost and operational requirements. NVIDIA is where our hands-on experience runs deepest.

Do you sell the hardware you recommend?

No. TensorFoundry is an advisory, assessment and optimisation practice. Purchasing stays with you and your chosen supplier.

What do we receive at the end?

A written architecture recommendation with preliminary capacity, memory, power and operational sizing, an indicative total cost of ownership, risks and a procurement roadmap, followed by a walkthrough with your team.

Does Discover lead into the other stages?

Only if you want it to. Each stage is independently purchasable. If you go on to buy hardware, Validate can independently assess it once your integrator has deployed it.

Planning a GPU purchase?

Start from your workloads, not a vendor's catalogue. NVIDIA, AMD and Apple Silicon assessed on their merits, including whether you need new hardware at all.

Discuss your workload Explore our services

Other stages

Three stages, each independently purchasable. Start where you are, or scope two or more together as one programme.