Field note 03 · September 2026
Cloud programs and investor readiness.
What the Google Cloud, NVIDIA, AWS, and Microsoft startup programs check, what pre-seed investors read on a deep-tech site, and how each is answered here — without overstating stage.
1 · Programs
What each program checks on the public site.
Common denominator: a real website on the company domain, an email on that domain, a clear statement that AI is the product, an incorporated entity, and a deck on request. This site supplies all five.
| Program | Tier / size (public terms) | Eligibility that touches the website | Where this site answers it |
|---|---|---|---|
| Google for Startups Cloud Program (AI program) | Start tier for pre-funding startups; Scale tier up to $200k over two years; AI-first Scale up to $350k. Credits cover Gemini and Gemma. | Public company website; company email domain matching the site; a billing account; founded within five years; Start needs a working MVP and business model; AI-first needs AI in the foundational structure; Scale needs public funding proof. | Product and Technology show AI is the product; contact@bytehub-ai.com on this domain; Company page states stage; Investors page carries funding proof later. |
| NVIDIA Inception (program) | Free; preferred hardware and software pricing, DLI training, VC Alliance, cloud-credit offers via partners (eligibility-based). | Working website, incorporation, under ten years old, at least one developer, a pitch deck upload, AI central to the product. | Technology names the Jetson stack we build on; Investors has the deck-request CTA; Company states incorporation. |
| AWS Activate | Founders tier for self-funded companies; Portfolio tier via a partner organization ID. | Active AWS account; partner ID for the higher tier. | Bedrock and AWS IoT named as a provider-aware cloud plane on Product and Investors. |
| Microsoft for Startups Founders Hub | Enrolled; a credit ladder unlocked by usage milestones. | Enrollment; workload milestones on the sponsored subscription. | Foundry named as a validation partner; the model-plane design is documented internally. |
2 · Investors
What a pre-seed investor reads on a deep-tech site, and where we put it.
Deep-tech pitching guidance is consistent: the cost of the problem, the moat, risk-reducing milestones, the team, and the ask — not specifications.
| Investor question | ByteHub's answer | Page |
|---|---|---|
| What breaks today, and what does it cost? | Offline is normal; data cannot leave; updates need proof. Teams hand-assemble local inference, cloud APIs, and OTA until a failed update or data-policy block stops the pilot. | Home, Investors |
| Why now? | Edge models are useful (1–9B at INT4); hybrid is a necessity (local autonomy plus cloud reasoning); model releases are becoming production and compliance risk (EU CRA). | Investors |
| What is the product, not the project? | EdgeRuntime, ModelOps control plane, policy-gated cloud intelligence, business sandbox; build → evaluate → sign → verify/run. | Product |
| Who buys first? | Industrial equipment OEMs, fleet operators, robotics and inspection integrators. Beachheads: private on-prem inspection AI; firmware and fleet-operations copilot. | Investors |
| Moat | Founder has operated the layers AI products ignore (i.MX95 bring-up, Yocto, kernel, U-Boot, RAUC, DDS); cross-silicon and provider-aware by design; firmware release discipline applied to models. | Company, Investors |
| Proof standard | Measured on our bench, then validated on a customer workflow, then published. No unmeasured numbers on the site. | Technology |
| Milestones | 0–3, 3–6, 6–12, 12–18 months, each with a product and a commercial milestone. | Roadmap |
| The ask and use of funds | Pre-seed SAFE; target, runway, initial close, illustrative cap, allocation — stated as in the deck, subject to discussion. | Investors |
3 · Cloud partners
What we build on, what each partner gains, what we ask.
Text only; names are platforms we build on and programs we are engaging, not endorsements.
| Partner | What ByteHub builds on | What the partner gains | Ask |
|---|---|---|---|
| Google Cloud | Vertex AI / Gemini for policy-authorized long-context reasoning, multimodal analysis, technical retrieval, evaluation. | Governed Gemini usage from real device fleets; a reference path for hybrid edge and cloud in industrial Linux. | AI-first credits, Vertex guidance, startup ecosystem. |
| NVIDIA | JetPack, CUDA, TensorRT / TensorRT-LLM / Edge-LLM; JetPack OTA extended with signed model artifacts; Orin and Thor on the bench. | Jetson becomes a deployable, auditable AI product platform. | Technical feedback on profiles and packaging, preferred Jetson pricing, DLI, Inception introductions. |
| AWS | Bedrock and Guardrails; AWS IoT identity and connectivity. | Governed Bedrock and IoT usage from edge fleets. | Activate credits, technical guidance. |
| Microsoft | Foundry (Azure-billed models), Entra, Key Vault; optional Arc / IoT Operations. | Governed Foundry usage; enterprise hybrid reference. | Credits held; co-sell path later. |
Programs and investors
Request the deck or start a program conversation.
The full deck, the technical brief, and the evaluation format are shared in a short conversation. No customer or NDA material is published here.