CoreWeave Startup Accelerator - GPU cloud credits + discounts
Source: https://www.coreweave.com/solutions/startups
Description
Create account to comment on specific lines or Sign in
+ 1 CoreWeave runs a Startup Accelerator Program that gives accepted companies CoreWeave Cloud credits plus additional discounts on GPU compute. The exact credit amount is not published — it is negotiated per-startup based on stage, workload, and consumption commitment. There are no application deadlines — the program is open-ended and rolling.
No comments on this line yet.
+ 2
No comments on this line yet.
+ 3 In 2025–2026 the program sits alongside two related vehicles: CoreWeave Ventures (cash + compute-for-equity investments, typically Seed / Series A) and partner funds like the NJEDA × CoreWeave $20M AI Hub Fund for New Jersey startups.
No comments on this line yet.
+ 4
No comments on this line yet.
+ 5 Bottom line for vibe coders / solo founders: This is not a self-serve free tier. You need an actual AI/ML/VFX/synthetic-bio company with a real workload — ideally already on a VC's radar — to get meaningful credits. Hobby projects and pre-product builders are better served by RunPod, Lambda, Modal, or the smaller per-hour GPU shops.
No comments on this line yet.
+ 6
No comments on this line yet.
+
7
No comments on this line yet.
+ 8
No comments on this line yet.
+ 10
No comments on this line yet.
+
11
1. Go to coreweave.com/solutions/startups (the canonical accelerator landing page — the older /coreweave-accelerator-program URL now 404s and redirects here).
No comments on this line yet.
+ 12 2. Click "Apply" / "Get Started" — it opens an intake form.
No comments on this line yet.
+ 13 3. Fill in company details: legal name, website, founder names, funding stage, total raised, lead investors, headcount.
No comments on this line yet.
+ 14 4. Describe your workload: model architecture / framework, target GPU SKU (A100 / H100 / H200 / GB200), number of GPUs, expected monthly GPU-hours, training vs inference, storage needs.
No comments on this line yet.
+ 15 5. Submit. Expect outreach from a CoreWeave startup / sales engineer rather than an automated approval — review is human and discretionary.
No comments on this line yet.
+ 16 6. If accepted you sign an order form / MSA that specifies the credit amount, validity window, and any required minimum commitment to unlock the discount tier.
No comments on this line yet.
+ 17
No comments on this line yet.
+ 18 Important caveats:
No comments on this line yet.
+ 19 • No published credit amount. Reports of "$5K–$200K" floating around online are anecdotal — do not assume a number until you have a signed order form.
No comments on this line yet.
+ 20 • Commitment-linked. Larger credit grants typically come with a usage commitment (similar to Render Scale or RunPod Growth tiers).
No comments on this line yet.
+ 21 • No credit card or instant signup — this is a B2B sales motion, not a self-serve trial.
No comments on this line yet.
+ 22 • Wait time: Plan for 1–3 weeks between submitting and getting an offer, longer if you don’t already have VC backing they recognize.
No comments on this line yet.
+ 23
No comments on this line yet.
+
24
No comments on this line yet.
+ 25
No comments on this line yet.
+ 27
No comments on this line yet.
+ 28 CoreWeave explicitly welcomes:
No comments on this line yet.
+ 29
No comments on this line yet.
+ 30 • AI / ML companies — the primary target. LLM training, fine-tuning, inference, RAG infra, agents, robotics models, foundation-model labs.
No comments on this line yet.
+ 31 • Visual effects & animation studios — GPU rendering, real-time VFX pipelines.
No comments on this line yet.
+ 32 • Synthetic biology — protein folding, molecular dynamics, drug-discovery models.
No comments on this line yet.
+ 33 • Other compute-intensive workloads — they say they will consider anything that genuinely needs lots of GPUs.
No comments on this line yet.
+ 34
No comments on this line yet.
+ 35 In practice, you will get further if you can show:
No comments on this line yet.
+ 36 • Real funding (pre-seed friendly companies are hard to land here — Ventures is the better door for those).
No comments on this line yet.
+ 37 • A named technical workload with a credible GPU-hour estimate.
No comments on this line yet.
+ 38 • A website + LinkedIn presence for the company.
No comments on this line yet.
+ 39
No comments on this line yet.
+
40
No comments on this line yet.
+ 41
No comments on this line yet.
+ 43
No comments on this line yet.
+ 44 If you’re raising or have raised a Seed / Series A, also consider CoreWeave Ventures — launched in 2025 as a dedicated venture arm that does:
No comments on this line yet.
+ 45
No comments on this line yet.
+ 46 • Cash investments at Seed / Series A (occasionally pre-seed or later).
No comments on this line yet.
+ 47 • Compute-for-equity — instead of (or in addition to) cash, they take equity in exchange for CoreWeave credits.
No comments on this line yet.
+ 48 • Production-grade testing environments, GTM intros to their enterprise customer base.
No comments on this line yet.
+ 49
No comments on this line yet.
+ 50 This is the main 2025–2026 evolution of the program and is now the primary path for VC-backed AI startups. Apply at coreweave.com/ventures.
No comments on this line yet.
+ 51
No comments on this line yet.
+
52
No comments on this line yet.
+ 53
No comments on this line yet.
+ 55
No comments on this line yet.
+ 56 Accepted accelerator startups can deploy on CoreWeave’s full fleet. As of May 2026 this includes:
No comments on this line yet.
+ 57
No comments on this line yet.
+ 58 GPU SKUMemoryInterconnectTypical use
No comments on this line yet.
+ 59 NVIDIA A10040/80 GB HBM2eNVLink + IBTraining, large-batch inference (legacy)
No comments on this line yet.
+ 60 NVIDIA A40 / A5000 / A600024–48 GBPCIe / NVLinkInference, rendering, smaller training
No comments on this line yet.
+ 61 NVIDIA L40 / L40S48 GBPCIeInference, generative video, rendering
No comments on this line yet.
+ 62 NVIDIA H100 (HGX)80 GB HBM3NVLink + 400G NDR InfiniBandLLM training & high-throughput inference
No comments on this line yet.
+ 63 NVIDIA H200 (HGX)141 GB HBM3eNVLink + 400G NDR InfiniBandLong-context LLMs, large MoE training
No comments on this line yet.
+ 64 NVIDIA GH20096 GB HBM3 + 480 GB LPDDR5XNVLink-C2CMemory-heavy inference, graph workloads
No comments on this line yet.
+ 65 NVIDIA RTX PRO 6000 Blackwell Server96 GBPCIeInference, content creation
No comments on this line yet.
+ 66 NVIDIA B200 / B300 (Blackwell, HGX)192–288 GB HBM3eNVLink + IBFrontier training
No comments on this line yet.
+ 67 NVIDIA GB200 / GB300 NVL7213 TB GPU mem per rack5th-gen NVLink (130 TB/s), Quantum-X / Spectrum-XFrontier-scale training clusters
No comments on this line yet.
+ 68
No comments on this line yet.
+ 69 Original accelerator launch (2022) advertised "more than 10 NVIDIA GPU SKUs — including A100s, A5000s and soon, H100s." That fleet has since expanded to the full Blackwell generation. Source: CoreWeave docs — GPU instances.
No comments on this line yet.
+ 70
No comments on this line yet.
+
71
No comments on this line yet.
+ 72
No comments on this line yet.
+ 74
No comments on this line yet.
+ 75 CoreWeave operates 30+ data centers across the US and Europe (London, Barcelona, Stockholm). Specific GPU SKUs are not available in every region — the latest hardware (GB200/GB300, B200) lands in primary US sites first. Confirm region availability with your CoreWeave SE before designing for a specific geography.
No comments on this line yet.
+ 76
No comments on this line yet.
+
77
No comments on this line yet.
+ 78
No comments on this line yet.
+ 80
No comments on this line yet.
+ 81 • NJEDA × CoreWeave AI Hub Fund — $20M fund (announced Dec 2025) for NJ-headquartered AI startups, providing capital plus CoreWeave compute access.
No comments on this line yet.
+ 82 • CoreWeave Ventures — see section above.
No comments on this line yet.
+ 83 • NVIDIA Inception — separate program from NVIDIA itself; CoreWeave is one of the cloud destinations Inception companies often land on.
No comments on this line yet.
+ 84
No comments on this line yet.
+
85
No comments on this line yet.
+ 86
No comments on this line yet.
+ 88
No comments on this line yet.
+ 89 • Don’t lead with "how much can I get for free?" — lead with the workload. The credit amount comes after a technical scoping call.
No comments on this line yet.
+ 90 • CoreWeave is enterprise-grade — minimum useful unit is typically a multi-GPU node, not a single GPU. If you need 1× H100 for a weekend project, use RunPod or Lambda.
No comments on this line yet.
+ 91 • Reserved-vs-on-demand pricing matters — once credits run out, you’ll convert to a reserved contract with a multi-month commit. Plan exit before signing.
No comments on this line yet.
+ 92 • VFX / animation / synth-bio applicants are real — this isn’t marketing fluff. CoreWeave hosts large render farms; mention that workload directly if it applies.
No comments on this line yet.
+ 93 • Existing CoreWeave customers generally don’t qualify for the accelerator credits — it’s aimed at new logos.
No comments on this line yet.
+ 94
No comments on this line yet.
+
95
No comments on this line yet.
+ 96
No comments on this line yet.
+ 97 Sources:
No comments on this line yet.
+ 98 • CoreWeave Solutions for Startups
No comments on this line yet.
+ 99 • CoreWeave Launches Startup Accelerator Program (blog, 2022)
No comments on this line yet.
+ 100 • CoreWeave Ventures
No comments on this line yet.
+ 101 • CoreWeave Investor News — Ventures Group launch (2025)
No comments on this line yet.
+ 102 • NJEDA × CoreWeave $20M AI Hub Fund
No comments on this line yet.
+ 103 • CoreWeave GPU Instance Documentation
No comments on this line yet.
+ 104 • CoreWeave Pricing
No comments on this line yet.
+ 105 • CoreWeave HGX H100/H200 product page
No comments on this line yet.