CoreWeave Posts Record Q2 Earnings and Defies GPU Obsolescence with A100 Contracts Through 2029
The massive financial engine powering AI infrastructure has posted record-breaking results, while simultaneously offering critical evidence to settle one of the most contentious debates of the artificial intelligence boom: the depreciation and obsolescence rate of AI silicon. On August 11, 2026, specialized AI cloud provider CoreWeave (CRWV) reported its second-quarter 2026 financial results, revealing that quarterly revenue more than doubled year-over-year to $2.575 billion (up 112.32% from $1.2 billion in Q2 2025).
CoreWeave's financial momentum is backed by a staggering $104.2 billion revenue backlog, which expanded by $25 billion in new contracts in just the five weeks following the end of the quarter. Pushed by insatiable demand for AI compute capacity, CoreWeave successfully implemented a 25% price increase across its SKUs in July 2026. The company raised its full-year 2026 revenue guidance to between $12.4 billion and $13.2 billion, up from its previous projection of $12 billion to $13 billion.
Defying the Obsolescence Narrative
Crucially, CoreWeave Chief Financial Officer Nitin Agrawal revealed to analysts that the company has signed a lucrative contract to rent out 2020-era Nvidia A100 GPUs well into 2029 at "full freight" and highly attractive prices.
This contract directly challenges the prevailing market fear that rapid hardware iterations—such as the transition from Hopper (H100/H200) to Blackwell (B200) and Rubin architectures—would render older-generation GPUs completely obsolete and trigger massive depreciation write-downs for data center operators. Instead, the deal demonstrates that older-generation chips remain highly useful and commercially viable for long-term inference and fine-tuning workloads.
Market Volatility and Overspend Fears
Despite these blockbuster operational metrics, neocloud and AI infrastructure stocks experienced sharp sell-offs on August 18, 2026. CoreWeave shares plummeted 7%, while competitors Nebius and Terawulf dropped 5% and 6% respectively. The sell-off reflects growing Wall Street anxiety regarding capex overspend and the long-term sustainability of the AI infrastructure buildout, highlighting a stark divergence between robust near-term business momentum and macroeconomic bubble fears.