Up to $100,000 in AWS credits, up to $150,000 in Nebius credits, plus $10,000 in NVIDIA training credits. That’s the headline for the NVIDIA Inception credits package, and it’s one of the more practical “free credits” bundles for AI startups right now.
Founders trying to stretch runway, ML engineers who need GPU access without committing to a big contract, and small teams shipping an AI product (not a slide deck) tend to get the most out of it. It’s also friendly to pre-revenue startups, which is honestly refreshing.
This guide breaks down NVIDIA Inception eligibility, the exact application steps, what the benefits actually include, and the catches that surprise people.
Program at a Glance
| Provider | NVIDIA |
| Credit Amount | Up to $100K AWS + up to $150K Nebius + more |
| Duration | No term limit; update profile every 6 months |
| Eligibility | AI/ML startups under 10 years; non-public; not excluded types |
| Credit Card Required? | No for Inception application; partner credits vary |
| Difficulty | Intermediate; form + pitch deck, then partner approvals |
| Best For | Startup cloud credits, DLI training, GPU discounts |
| Official Page | NVIDIA Program Page |
What You Actually Get
NVIDIA Inception is a free, virtual accelerator program for AI startups with no equity, no fees, and no cohort deadlines. Once you’re in, you get access to a benefits portal where you can request partner cloud credits (not issued directly by NVIDIA), including up to $100,000 via AWS Activate and up to $150,000 via Nebius AI Lift. You also get $10,000 in NVIDIA Deep Learning Institute (DLI) training credits, 30% off NVIDIA DGX Cloud, and preferred pricing on select NVIDIA GPUs (up to about 30% off) through authorized resale partners. On top of the credits, there’s a real go-to-market layer: co-marketing assets, Inception badges, and an eligibility-based VC intro path through Inception Capital Connect.
In real terms, the value usually lands in two buckets. First is compute you would have paid for anyway (cloud credits and GPU discounts). Second is leverage: DLI training + platform support that helps a small team move faster, especially if you’re hiring or ramping new engineers. The “up to” amounts are big, but you should plan like you’ll need to qualify for each partner tier rather than assuming the max shows up automatically.
Who Qualifies (and Who Doesn’t)
NVIDIA Inception is intended for legitimate AI/ML startups where AI is central to the product. NVIDIA also has explicit exclusions and a company age cap, so it’s worth checking those before you spend time polishing your deck.
- Your company must be under 10 years old, since older companies aren’t eligible.
- You’ll need a pitch deck (PDF) that clearly explains mission, product, team, and traction.
- The application must be submitted in English, which catches some teams off guard.
- You have to keep your Inception profile updated every 6 months to keep access to benefits.
If you’re a public company, a reseller/distributor, a cloud service provider, a consulting/outsourced dev shop, or a crypto-focused company, you do not qualify. Cryptocurrency-focused startups are explicitly excluded (NVIDIA has enforced this since 2025).
How to Sign Up
Plan for about 30 minutes if your pitch deck is ready.
- Go to nvidia.com/en-us/startups.
- Click “Join Inception” or “Apply Now”.
- Fill out the application form. You will provide your company name, website URL, incorporation date, a description of your AI/ML product, funding stage, team size, competitors, and mission.
- Upload a pitch deck (PDF) covering mission, product, team, and traction so NVIDIA can determine how to support you.
- Submit the application (it must be in English).
After submission, NVIDIA reviews the application with no public timeline; expect roughly a couple weeks. Once approved, you’ll get access to a personalized Inception portal where you can request benefits like cloud credits and DLI training codes.
What the Credits Cover
The key thing to understand is that NVIDIA Inception benefits are a bundle of partner programs and discounts, not one single “NVIDIA wallet.” You apply once to Inception, then request specific benefits (AWS Activate, Nebius AI Lift, DLI codes, and so on) inside the portal.
| Service / Feature | What It Does | Included? |
|---|---|---|
| AWS Activate credits | Cloud credits to run workloads on AWS, including NVIDIA GPUs on EC2. | Partial |
| Nebius AI Lift credits | Cloud credits (including inference credits) with early access to Blackwell GPUs. | Partial |
| DLI training credits | $10K in credits for self-paced courses, workshops, and certifications (with GPU labs). | ✓ |
| DGX Cloud discount | 30% off DGX Cloud, with a minimum commitment and minimum spend. | ✓ |
Notable exclusions: the cloud credits are not automatic, and they are not paid directly by NVIDIA. Also, the DGX Cloud discount exists, but the minimum spend is high, so it’s not a “free tier” substitute for most early teams.
Limitations to Know About
Every free credit program has catches. NVIDIA Inception’s are mostly about partner approvals, ongoing eligibility, and minimum commitments.
- AWS credits are not guaranteed even after you’re accepted into NVIDIA Inception, and some startups report inconsistent approval outcomes.
- The AWS $100K tier may informally expect about $250K+ in institutional funding, though it’s applied inconsistently.
- You must update your Inception profile every 6 months or you can lose access to benefits.
- DGX Cloud’s 30% discount comes with a minimum commitment of 4 nodes for 1 month and a minimum spend of $75,000.
When credits run out, you’re not automatically “kicked out” of Inception. But the underlying partner services behave like normal paid cloud services: if you keep using them, you should expect to pay under that provider’s billing terms. Put billing alerts in place before you request anything sizable, especially if you’re stacking multiple partner programs.
Have Unused NVIDIA Credits?
It’s common for startups to get access to more credits than they can realistically burn before a program window closes or strategy changes. Maybe you moved off a provider, maybe you raised later than expected, maybe the team just didn’t need that much compute. If you’re sitting on unused NVIDIA-related credits or allocations you can’t use, AI Credit Mart lets you sell unused credits instead of watching them expire.
Need More NVIDIA Credits?
Once your free credits are exhausted (or if you never got approved for the full partner tiers), paying retail is not the only option. AI Credit Mart lists discounted NVIDIA credits and related allocations from teams that can’t use everything they received. Discounts typically land around 30–70% below retail, depending on what’s available and how soon it expires.
Tips for Getting the Most Out of Your Credits
- Make AI/ML central in your pitch deck, because NVIDIA wants to see it’s core to the product.
- Apply for AWS credits early, and use the same email for your Inception account and your AWS account.
- If AWS rejects your credits request, try again or escalate through the NVIDIA Developer Forums instead of assuming it’s final.
- Don’t ignore profile maintenance. Put a calendar reminder to update your Inception profile every 6 months.
- Stack partner programs when it makes sense: AWS Activate, Nebius AI Lift, FluidStack, Scaleway, and Azure can be combined in practice.
Frequently Asked Questions
Potentially a few hundred thousand dollars in combined value: up to $100,000 via AWS Activate, up to $150,000 via Nebius AI Lift, up to $100,000 via FluidStack (CPU credits), plus $10,000 in DLI training credits and a 30% DGX Cloud discount. The exact value you realize depends on which partner credits you’re approved for and whether you can actually use them before they become irrelevant to your stack. Practically, teams get the most “real” value from the credits they can activate quickly (training + at least one cloud partner). Treat the maximum numbers as targets, not guarantees.
No for the Inception application itself; partner programs may have their own billing requirements.
Inception membership has no term limit, but you must update your profile every 6 months to stay eligible for benefits.
Yes. If you have NVIDIA credits you won’t use before they expire, you can list them on AI Credit Mart and sell them at up to 70% of face value. Companies regularly list surplus credits from startup programs and enterprise agreements.
AI Credit Mart has discounted NVIDIA credits available from companies with surplus allocations. Prices are typically 30-70% below retail.
They generally become unusable, and if you keep consuming the underlying service you should expect normal paid billing from that provider.
Because AWS evaluates the Activate request separately from NVIDIA Inception approval, and AWS may reject applications even after you’re accepted into Inception. Some founders also report that higher tiers can depend on factors like funding level (the $100K tier is sometimes associated with about $250K+ institutional funding), but the enforcement is inconsistent. The best move is to apply early, use the same email across accounts, and re-apply or escalate via NVIDIA Developer Forums if you get denied.
Usually no, since the minimum spend is $75,000 with a 4-node, 1-month minimum commitment.
NVIDIA Inception is a high-upside program if you’re building a real AI product and you’re willing to chase partner benefits through the portal. Get in, request credits early, and keep your profile current so the good stuff doesn’t quietly disappear.
Your AI credits are losing value every day
Join the marketplace and start trading unused credits today.