Skip to content
Shared AI Researchan open archive

AI hardware23 accelerators

Accelerator generations and prices

The chips the era is built on, by vendor and generation, with indicative unit prices and cloud rates. Vendors rarely publish list prices; every figure here is approximate.

Data compiled as of July 2026. Corrections by pull request are welcome.

NVIDIA8

ChipYearMemoryBandwidthComputeUnit price (approx.)Cloud rate (approx.)
V100 (Volta)
Introduced tensor cores; trained the GPT-2/GPT-3-era models.
201716–32 GB HBM2900 GB/s125 TFLOPS FP16~$10,000~$0.30–2/hr
A100 (Ampere)
The workhorse of the ChatGPT moment; GPT-4 reportedly trained on ~25k A100s.
202040–80 GB HBM2e2.0 TB/s312 TFLOPS BF16~$10,000–15,000~$0.75–3/hr
H100 (Hopper)
The scarcest commodity of 2023–24; secondary-market prices have since fallen sharply as Blackwell ramped.
202280 GB HBM33.35 TB/s~1.0 PFLOPS FP8~$25,000–40,000 new; used units now ~$10,000–20,000~$2–8/hr
H200
Memory-expanded Hopper; nearly doubled inference throughput on large models.
2024141 GB HBM3e4.8 TB/s~1.0 PFLOPS FP8~$30,000–40,000~$3.50–9/hr
B200 (Blackwell)
Dual-die design with FP4 inference; sold primarily as GB200 NVL72 racks (~$3–3.4M each).
2024180–192 GB HBM3e8 TB/s~4.5 PFLOPS FP8 / 9 PFLOPS FP4~$30,000–40,000~$4–12/hr
GB300 / B300 (Blackwell Ultra)
Mid-cycle refresh for reasoning-model inference; now the volume Blackwell part as Rubin ramps.
2025288 GB HBM3e8 TB/s~1.5× B200 (FP4)rack-scale (~$3.7–4M per NVL72; ~$40–55k per GPU)~$5–18/hr
VR200 (Vera Rubin)
Sampling now, volume H2 2026; HBM4 costs (~$500+ per stack) pushed rack prices to roughly double GB300.
2026288 GB HBM4~13 TB/s~50 PFLOPS FP4 (per dual-die package)rack-scale (~$7.8–8.8M per NVL72; ~$55k per GPU est.)
Groq 3 LPX (LPU)
Product of NVIDIA’s ~$20B Groq licensing acquihire (Dec 2025); deterministic low-latency decode engine paired with Rubin racks.
2026500 MB SRAM per LPU150 TB/s (on-chip SRAM)~315 PFLOPS per 256-LPU rackrack-scale; pricing not yet public

Google4

ChipYearMemoryBandwidthComputeUnit price (approx.)Cloud rate (approx.)
TPU v4
Optical circuit switching across 4,096-chip pods; trained PaLM.
202132 GB HBM21.2 TB/s275 TFLOPS BF16not sold; cloud only~$2–3/hr
TPU v5e / v5p
Split the line into efficiency (v5e) and performance (v5p) tiers; trained Gemini 1.0.
202316 / 95 GB HBM0.8 / 2.8 TB/s197 / 459 TFLOPS BF16not sold; cloud only~$1.20 / $4.20/hr
TPU v6 (Trillium)
4.7× v5e per-chip performance; trained Gemini 2.x.
202432 GB HBM1.6 TB/s~926 TFLOPS BF16not sold; cloud only~$2.70/hr
TPU v7 (Ironwood)
GA on Google Cloud April 2026; Anthropic contracted for up to ~1M chips. TPU v8 splits into Sunfish (training) and Zebrafish (inference) for ~2027.
2025192 GB HBM3e7.4 TB/s~4.6 PFLOPS FP8not sold; cloud only~$5.40–12/hr (commit vs. on-demand; large contracts far lower)

AMD4

ChipYearMemoryBandwidthComputeUnit price (approx.)Cloud rate (approx.)
MI250X (CDNA 2)
Powered Frontier, the first exascale supercomputer; little LLM traction.
2021128 GB HBM2e3.2 TB/s383 TFLOPS FP16~$12,000–15,000
MI300X (CDNA 3)
AMD’s first credible H100 rival; adopted by Microsoft, Meta, and OpenAI for inference.
2023192 GB HBM35.3 TB/s~1.3 PFLOPS FP8~$15,000–20,000~$2–4/hr
MI355X (CDNA 4)
Anchor of OpenAI’s 6-gigawatt AMD partnership announced October 2025.
2025288 GB HBM3e8 TB/s~5 PFLOPS FP8 / 10 PFLOPS FP4~$25,000–30,000~$2.60–8.60/hr
MI455X (CDNA 5) / Helios
AMD’s NVL72 answer: 72 GPUs, 31 TB HBM4, open Ethernet fabric; claims ~30% more tokens per dollar than Rubin. Ships late 2026.
2026432 GB HBM4~19.6 TB/s~40 PFLOPS FP4 / 20 PFLOPS FP8rack-scale (~$5–5.5M per 72-GPU Helios rack)

Amazon2

ChipYearMemoryBandwidthComputeUnit price (approx.)Cloud rate (approx.)
Trainium 2
Anthropic’s Project Rainier cluster runs on hundreds of thousands of these chips.
202496 GB HBM32.9 TB/s~650 TFLOPS FP8not sold; cloud only~$1–2/hr (per chip, Trn2 instances)
Trainium 3
AWS’s first 3nm chip, GA December 2025; UltraServers scale to 144 chips, with Anthropic and Bedrock running production workloads.
2025144 GB HBM3e4.9 TB/s~2.5 PFLOPS FP8not sold; cloud only~$1.80/hr (per chip, est.)

Microsoft1

ChipYearMemoryBandwidthComputeUnit price (approx.)Cloud rate (approx.)
Maia 200
Microsoft’s first competitive inference silicon (Jan 2026); claims ~3× the FP4 of Trainium 3 or Ironwood per chip.
2026216 GB HBM3e7 TB/s~5 PFLOPS FP8 / 10 PFLOPS FP4not sold; internal Azure only

Meta1

ChipYearMemoryBandwidthComputeUnit price (approx.)Cloud rate (approx.)
MTIA v2 / "Iris"
Runs Meta’s ads and recommendation inference; the GenAI-focused "Iris" chip ramps late 2026 on a roughly six-month silicon cadence.
2026128 GB LPDDR5 (v2)~0.2 TB/s (off-chip, v2)~354 TOPS INT8 (v2)not sold; internal only

Intel1

ChipYearMemoryBandwidthComputeUnit price (approx.)Cloud rate (approx.)
Gaudi 3
Priced aggressively but found little adoption; Intel shelved Falcon Shores and pivoted to the Crescent Island inference GPU, sampling H2 2026.
2024128 GB HBM2e3.7 TB/s~1.8 PFLOPS FP8~$16,000 (8-chip board ~$125k)

Cerebras1

ChipYearMemoryBandwidthComputeUnit price (approx.)Cloud rate (approx.)
WSE-3
A single wafer-scale chip; IPO’d May 2026 (~$23B target) on the back of a ~$20B, 750 MW OpenAI inference agreement.
202444 GB on-wafer SRAM21 PB/s (on-chip)~125 PFLOPS FP16 (sparse)~$2–3M per CS-3 system

SambaNova1

ChipYearMemoryBandwidthComputeUnit price (approx.)Cloud rate (approx.)
SN40L (RDU)
Dataflow architecture for trillion-parameter inference; JPMorgan deployment and a $350M Series E (Feb 2026) after Intel acquisition talks collapsed.
202364 GB HBM + 1.5 TB DDR (three-tier)~1.6 TB/s (HBM)688 TFLOPS BF16sold as full systems/cloud; pricing not public

Prices are indicative street or list prices per accelerator at launch in USD. Real transactions vary widely with volume, form factor, and packaging; rack-scale systems are priced as systems, not chips. Cloud rates are typical on-demand per-accelerator-hour figures across major and specialist providers.