Amazon: AI deals, investors and valuation

At the Invest in Germany Summit Amazon announced over €16B in Germany for 2027, €4.8B of it for the AWS Frankfurt region.

Qualcomm co-develops custom inference chips and 1.6 Tbps optical interconnects for AWS; $60 billion is the purchase ceiling at which the warrant fully vests.

Qualcomm issues Amazon a warrant for up to 25 million shares at $161.26, vesting in tranches with purchases.

Riding demand, Amazon adds 2 million Nvidia GPUs (Blackwell Ultra, Rubin, Rubin Ultra) for AWS, with deliveries in 2027-2028.

Microsoft, Google, Nvidia and Amazon have offered to directly finance new SK Hynix production lines and tools to secure memory supply.

Meta signs a multi-year, multibillion-dollar contract to use tens of millions of AWS Graviton5 CPU cores for AI agents.

$5 billion immediately (at a $380 billion valuation) and up to $20 billion tied to commercial milestones, on top of the $8 billion already invested.

In exchange for Amazon's investment, Anthropic commits to spend over $100 billion on AWS over 10 years, up to 5 GW (Trainium2/3).

At GTC, AWS announces more than 1 million Nvidia Blackwell and Rubin GPUs from 2026 to 2027, plus ConnectX and Spectrum-X networking.

Amazon raises the equivalent of $53.8 billion ($37 billion in 11 tranches and €14.5 billion in its euro debut), the largest corporate offering in history, to fund AI.

At MWC Amazon added €18B for Spanish data centres in Aragon, lifting its Spanish total to €33.7B - its largest investment outside the US.

Amazon leads the $110 billion round ($730 billion pre-money) with $50 billion: $15 billion upfront and $35 billion conditional. Completed per the SEC filing of 31/07/2026; AWS becomes the exclusive third-party cloud for the Frontier platform.

The $38 billion AWS contract grows by $100 billion over eight years; OpenAI commits to use at least 2 GW of Trainium chips and brings its models to Bedrock.

Amazon announced over $35B in India by 2030 across cloud, AI, logistics and exports, on top of ~$40B already invested; the AI share is not broken out.

AWS and HUMAIN expand their deal: up to 150,000 accelerators (Nvidia GB300 and Trainium) in Riyadh's AI Zone, AWS becoming HUMAIN's preferred AI partner.

First major contract with AWS: $38 billion over seven years for hundreds of thousands of Nvidia GPUs (GB200/GB300) on EC2 UltraServers.

Amazon Web Services signs a 15-year contract worth about $5.5 billion for AI space and power at Cipher's data centers, delivery in 2026.

AWS activates Project Rainier, a cluster of about 500,000 Trainium2 chips dedicated to Anthropic, which aims for over 1 million chips by end-2025.

At APEC, AWS pledged over $5bn by 2031 for new AI data centers in Incheon and Gyeonggi, on top of its Ulsan campus with SK.

AWS and SK Group are building a 103MW AI campus in Ulsan (expandable to 1GW), run by SK Telecom; total ≈7tn won, about $4bn from AWS.

With PM Albanese, AWS announced A$20B over 2025-2029 for Sydney and Melbourne data centres plus three solar farms.

17-year contract for power from the Susquehanna plant, destined for AWS data centers in Pennsylvania.

AWS launched its Taipei cloud region and pledged over $5bn to build and run data centers; customers include TSMC.

AWS and HUMAIN jointly invest more than $5B in a Saudi 'AI Zone' with dedicated AWS infrastructure, UltraClusters and Bedrock.

AWS will invest $4B in its first data centres and cloud region in Chile, operational from H2 2026.

At Davos AWS signed an MOU with Maharashtra for $8.3B by 2030 to expand its Mumbai cloud region with new data centres and AI services.

Multi-year, multi-generation agreement: Marvell supplies AWS with custom Trainium 2 chips, optical DSPs, interconnect modules and Ethernet switches.

Second $4 billion investment; Anthropic picks AWS as primary training partner and Trainium chips.

Amazon's first investment in Anthropic: $1.25 billion upfront and $2.75 billion completed in March 2024. AWS becomes primary cloud provider.