I’m 18 and Got $300K in Free Startup Credits for My AI Company. Here’s the Exact Playbook.

What NVIDIA, Google, AWS and Microsoft actually give, who qualifies, and the small mistakes that get founders rejected.

The first $300K that went into my startup didn’t come from an investor.

It came from NVIDIA, Google, AWS and Microsoft, in the form of startup credits. No equity. No board seat. No pitch meeting.

I’m 18. I run Sumora, where we’re building “Shopify for AI agents”: one operating system where agents run outbound, marketing, engineering and support for a business. Our outbound agent is live with 6 customers paying $500 a month.

And right now, our expected infrastructure bill is $0 until we reach roughly 500 paid users.

A lot of founders either don’t know these programs exist, or apply once, get rejected for a reason they never understood, and give up. So this is the exact playbook: what each program actually gives, who qualifies, and the small mistakes that quietly get applications denied.

Why credits are the most underrated funding for AI founders

For an AI startup, compute isn’t a line item. It’s the business. Every agent run, every model call and every database query shows up on a cloud bill.

That makes credits unusually valuable for AI founders specifically. A software startup saves on hosting. An AI startup can cover the single biggest cost of delivering its product.

Credits buy you time to find product-market fit without giving away equity to pay for servers.

Two caveats before the playbook. Credits are not cash, so they won’t pay salaries or rent. And they expire, so the goal isn’t to collect the biggest number. It’s to line up credits with the months you’ll actually need them.

Start with NVIDIA Inception

My first approval came from NVIDIA Inception. I didn’t have a fancy deck or a warm intro. All I showed them was the product, which had roughly 500 users at the time.

That’s the part people overthink. Inception is free and takes no equity. According to NVIDIA’s FAQ, you need four things:

1. At least one developer on the team
2. A working website
3. An incorporated company
4. A company less than 10 years old

There’s no revenue requirement, no funding requirement and no cohort deadline. You don’t even need to use NVIDIA GPUs yet.

The benefits are broader than most people expect: free cloud credits from NVIDIA and its partners, free self-paced courses through the Deep Learning Institute, preferred pricing on select hardware and software, and access to NVIDIA’s VC Alliance.

Why start here? Because it’s the easiest real “yes” to get, and an Inception badge on your site and deck makes you look more serious to every program that comes after it.

The first approval is the hardest. Every one after it is easier.

Google for Startups Cloud: the biggest number, the strictest rules

Google has the largest headline figure, and it’s worth understanding the tiers before you apply, because applying to the wrong one is the most common way to get rejected.

The Google for Startups Cloud Program lists three tiers:

• Pre-funded: $2,000 in credits to build and iterate on an MVP
• Early stage (Seed to Series A): $200,000, or up to $350,000 for AI-first startups
• Series B and later: customized support, with no fixed credit amount listed

The AI track is where it gets interesting. Per Google’s AI program page, AI-first startups can get 100% of up to $250,000 in year one, plus 20% of up to another $100,000 in year two.

But read the eligibility carefully, because this is where founders trip:

• You need a Pre-Seed or Seed round in the last 5 years, or a Series A in the last 12 months.
• The money has to come from institutional investors or VCs. SAFEs count. Angels, friends and family, grants, prize money and crowdfunding do not.
• You can’t have already received more than $5,000 in Google Cloud credits.
• Your product needs to use, or plan to use, Gemini as a foundation of what you sell.

And one gotcha almost nobody mentions: Google says a company domain that subscribed to a paid Google Workspace plan within 31 days before applying may be disqualified from that benefit. If you just set up paid Workspace for your team, check the timing before you hit submit.

If you’ve only raised from angels or not at all, apply to the pre-funded tier, not the big one.

AWS and Microsoft: small first, big through a provider

AWS Activate has two doors, and the difference between them is the most useful thing to know.

Activate Founders gives $1,000 in credits. You qualify if the company was founded in the last 10 years, is self-funded or pre-Series B, and has a fully working website.

Activate Portfolio goes up to $100,000. The catch is that you need an Organizational ID from an Activate Provider, which typically means an accelerator, incubator or investor in AWS’s network. If you’re in one, ask them for it before applying through the public form.

A few details that decide approvals:

• You apply with an AWS Builder ID tied to a professional email, not a Gmail address.
• AWS says inaccurate startup information can get an application denied, so your website, company details and stage need to match.
• Decisions typically arrive in 7 to 10 business days.

One more reason AWS credits matter for AI startups: they can be used on third-party foundation models in Amazon Bedrock.

The $1,000 tier is a door, not a destination. The real credits come through a provider.

And Microsoft for Startups

Microsoft for Startups advertises up to $150,000 in credits that can be applied across eligible Azure services.

Microsoft’s public page doesn’t spell out the tiers or eligibility, and the program has changed several times over the years. So my advice is simple: sign up through the portal, see what your startup unlocks today, and don’t plan your runway around a number you read in an old blog post.

Timing matters more than the number

The biggest mistake isn’t getting rejected. It’s getting approved too early.

Credits come with start dates and expiry dates. If you activate a large package while you’re still figuring out what to build, a chunk of it can expire before you have real usage to spend it on.

Google’s AI tier makes the intended rhythm visible: most of the value lands in year one, with a smaller top-up in year two. That’s designed for a company that’s already shipping and about to grow, not one still sketching ideas.

So I think about it in stages. Small starter credits while you build. Bigger packages right before you expect usage to climb, like a launch, a paid pilot or the moment your first agent goes live for customers.

Apply for the big credits when you’re about to need them, not the day you incorporate.

The playbook, in order

If I were starting from zero tomorrow, this is the order I’d follow:

1. Get the basics right: incorporate, ship a working website, and set up an email on your own domain.
2. Ship something people use. Real usage beats a polished deck. Mine was a product with about 500 users.
3. Apply to NVIDIA Inception first. It’s free, fast to qualify for, and becomes social proof.
4. Take the starter tiers next: AWS Activate Founders and Google’s pre-funded tier.
5. Join an accelerator or program that’s an Activate Provider to unlock the larger AWS credits.
6. Apply to Google’s early-stage or AI tier only after you’ve raised from an institutional investor.
7. Track every credit’s amount, start date and expiry in one sheet, so nothing expires unused.

Notice what’s missing: a big fundraise. Most of this list is available to a student founder with a real product and a working website.

Investors fund your story. Credit programs fund your product. Get the second one first.

The fastest path to credits is a product that people already use.

What to put in the application

None of these forms ask for a pitch deck essay. They ask for a few facts, and reviewers are checking whether those facts add up.

Here’s what I’d make sure is true and visible before applying anywhere:

• A website that explains what the product does in one line, with a way to try it or join
• A company email on your own domain that matches the website
• Company details that match everywhere: name, incorporation date, country and stage
• Proof of use, like users, customers or a live demo, even if the numbers are small

The common thread is consistency. A reviewer who sees a Gmail address, a half-built landing page and a company name that doesn’t match the incorporation record has an easy reason to say no.

And don’t stop at the big four. Many developer tools, from analytics to databases to code sandboxes, run their own startup programs. Search “[tool name] for startups” for every service on your stack. Individually they’re small. Together they add up.

Can students apply?

It’s a fair question, so I’ll answer it directly.

None of the eligibility rules I’ve quoted here mention age or student status. What they ask for is a real company: incorporated, with a working website, a developer on the team and, for the bigger tiers, the right kind of funding.

So the blocker for most student founders isn’t being a student. It’s not having incorporated yet, or not having shipped anything people can use. Fix those two and you’re in the same position as any other early-stage founder.

Programs don’t fund ages. They fund companies with products.