Amazon is moving $8 billion worth of Nvidia Grace Blackwell AI chips into a special purpose vehicle that will sell bonds to investors and lease the hardware back. The deal keeps Amazon using the chips across at least five U.S. states while shifting the cost off its balance sheet. With quarterly capital spending at $54.2 billion and negative free cash flow of -$7.6 billion, the move helps protect Amazon's double-A credit rating. If it works, other Big Tech companies may adopt the same approach.
Investors should be cautious about depreciation risk. AI chips lose value quickly as new generations launch, and Nvidia is also building financing platforms with six financial partners to raise over $500 billion for AI infrastructure. The scale of these deals shows how central hardware ownership is to the current AI spending cycle.
Investor Steve Eisman warns that the AI industry leans heavily on just two customers, OpenAI and Anthropic. If either stumbles, the chain stretching from Nvidia to cloud providers like Microsoft, Amazon, and Google could break. He also notes that hyperscaler cash flow is shrinking and that AI funding may come mostly from tech giants rather than venture capital, comparing the concentration risk to the subprime housing crisis he predicted before 2008.
On the policy front, former Anthropic researcher Coxon will testify at a New York City AI hearing requested by City Council Speaker Julie Menin. Lawmakers are considering a package of bills designed to set safeguards around AI technology. The hearing signals growing local interest in regulating the industry from the ground up.
OpenAI President Greg Brockman and his wife Anna pulled a planned second $25 million donation to the Leading the Future super PAC. Brockman said the group became a distraction for OpenAI employees, and he informed the company internally in June that no further donations would be made. OpenAI stated it does not direct or oversee the super PAC's activities.
SoftBank founder Masayoshi Son is investing $4 billion in DigitalBridge, a U.S. data center firm. The deal will make DigitalBridge its third-party infrastructure partner and is one of the largest in SoftBank's history. Son aims to build a global technology empire that can rival companies like Amazon and Google while diversifying beyond telecom and finance.
Innodata opened a New Jersey R&D lab with motion-capture specialist Vicon, using sub-millimeter infrared technology to produce ultra high-precision 3D motion data. The facility serves humanoid and industrial robot training, offering off-the-shelf datasets, custom projects, and third-party performance validation. The lab deepens Innodata's role in physical AI infrastructure.
The ECB warns that AI risks are converging across financial markets. Nearly 90% of major euro-area banks use generative AI, and about 70% of EU securities firms plan to increase AI investment. Christine Lagarde cautioned that if banks and funds use similar models, they may make identical trades simultaneously, amplifying market shocks. Only 5% of asset managers currently allow autonomous AI trading, but that share is expected to grow.
A new guide lists 10 free generative AI courses from Google, Microsoft, IBM, AWS, NVIDIA, and Hugging Face. The programs cover large language models, AI agents, prompting, diffusion models, and responsible AI, ranging from beginner introductions to more technical content. Chennai government colleges are also seeking approval for new undergraduate courses in data science, AI, and defence studies ahead of the 2026-27 academic session.
Key Takeaways
- Amazon shifts $8 billion in Nvidia AI chips off its balance sheet through a sale-leaseback deal with a special purpose vehicle.
- Investors face depreciation risk as AI chips lose value quickly when new chip generations launch.
- Eisman warns AI industry depends on just two customers, OpenAI and Anthropic, creating concentration risk across the supply chain.
- SoftBank invests $4 billion in data center firm DigitalBridge to rival Amazon and Google in global infrastructure.
- Former Anthropic researcher Coxon will testify at a New York City AI hearing on proposed AI safeguards.
- OpenAI's Greg Brockman cancels a second $25 million donation to the Leading the Future super PAC, citing employee distraction.
- ECB warns nearly 90% of euro-area banks use generative AI, raising systemic risk if institutions trade in lockstep.
- Innodata opens a New Jersey motion-capture lab with Vicon for high-precision 3D training data for humanoid and industrial robots.
- Free generative AI courses from Google, Microsoft, IBM, AWS, NVIDIA, and Hugging Face cover topics from prompting to responsible AI.
- Chennai government colleges seek approval for new undergraduate AI, data science, and defence studies courses for 2026-27.
Amazon plans $8 billion Nvidia chip leaseback to ease AI spending
Amazon is moving $8 billion worth of Nvidia Grace Blackwell AI chips into a special purpose vehicle. The vehicle will sell bonds to investors and lease the chips back to Amazon. Amazon will keep using the chips in data centers across at least five U.S. states. The deal helps Amazon reduce the financial burden of AI hardware on its balance sheet. It also protects Amazon's double-A credit rating. If successful, other Big Tech companies may follow the same approach.
Amazon shifts $8 billion in Nvidia chips off its balance sheet
Amazon is restructuring ownership of $8 billion in Nvidia AI chips through a sale-leaseback deal. A special vehicle will buy the chips and sell debt to investors. Amazon will lease the hardware back and keep operating it. The chips are already installed in data centers across at least five states. Amazon spent $54.2 billion on capital projects in the second quarter. The company's free cash flow turned negative at -$7.6 billion over the past year. Investors face risk because AI chips lose value quickly as new generations launch.
Amazon moves $8B Nvidia chips off books through leaseback deal
Amazon is placing $8 billion in Nvidia AI chips into a special purpose vehicle. The vehicle will raise money from investors through debt and lease the chips back to Amazon. This shifts the cost of the hardware off Amazon's balance sheet while keeping operations unchanged. Investors face depreciation risk as new chip generations make older models less valuable. Nvidia is also building similar financing platforms with six financial partners to raise over $500 billion for AI infrastructure.
Innodata opens high-precision motion-capture lab for physical AI
Innodata opened a new R&D laboratory in New Jersey built with motion-capture specialist Vicon. The lab produces ultra high-precision 3D motion data for training humanoid and industrial robots. It offers both off-the-shelf and custom motion-capture datasets. The facility also provides third-party performance validation for physical AI systems. This move deepens Innodata's role in data infrastructure for robotics and trust-and-safety benchmarking.
Innodata launches robot data lab with Vicon motion capture
Innodata opened a New Jersey R&D lab equipped with sub-millimeter infrared motion capture technology developed with Vicon. The facility produces and validates high-precision 3D training data for humanoid and industrial robots. Clients can access off-the-shelf datasets, custom motion projects, and independent performance benchmarking. The lab positions Innodata further into physical AI infrastructure. Investors are watching how this shapes Innodata's investment narrative in the robotics data space.
Eisman warns AI bubble depends on just two customers
Investor Steve Eisman says the AI industry depends heavily on just two customers, OpenAI and Anthropic. If either company struggles, the entire chain from Nvidia to cloud providers like Microsoft, Amazon, and Google could break. He also warns that hyperscaler cash flow is shrinking and that AI funding may come mostly from tech giants rather than venture capital. Eisman compares the risk to the subprime housing crisis he predicted before 2008.
SoftBank invests $4 billion in data center firm DigitalBridge
SoftBank founder Masayoshi Son is investing $4 billion in DigitalBridge, a US company that specializes in data centers. The deal is one of the largest in SoftBank's history and will make DigitalBridge its third-party infrastructure partner. Son aims to build a global technology empire that can rival companies like Amazon and Google. The investment also helps SoftBank diversify beyond its traditional telecom and finance businesses.
Former Anthropic researcher to testify at New York City AI hearing
Former Anthropic researcher Coxon will testify at a New York City AI hearing. The appearance was requested by City Council Speaker Julie Menin, who has encouraged AI whistleblowers to speak out. The hearing takes place as city lawmakers consider a package of bills designed to set safeguards around AI technology. The report was shared by Bloomberg News on October 4.
OpenAI's Greg Brockman cancels second $25 million donation to AI Super PAC
OpenAI President Greg Brockman and his wife Anna pulled a planned second $25 million donation to the Leading the Future super PAC. Brockman said he felt terrible that the group became a distraction for OpenAI employees. He informed the company internally in June that no further donations would be made beyond the initial $25 million. OpenAI also stated in a blog post that it does not direct or oversee the super PAC's activities.
10 free generative AI courses to take online in 2026
A new guide lists 10 free generative AI courses from companies like Google, Microsoft, IBM, AWS, NVIDIA, and Hugging Face. The courses cover topics including large language models, AI agents, prompting, diffusion models, and responsible AI. Options range from beginner-friendly introductions to more technical programs for experienced learners. Some courses offer digital credentials, though certain features like certificates may require paid plans.
ECB warns AI risks are converging across financial markets
ECB President Christine Lagarde stated that nearly 90% of major euro-area banks use generative AI. About 70% of EU securities firms plan to increase AI investment. She warned that if banks and funds use similar AI models, they may make the same trades at the same time. This could amplify market shocks instead of spreading risk. Only 5% of asset managers currently allow AI to make autonomous trade decisions, but that number is expected to grow.
Chennai government colleges seek new AI and data science courses
Chennai city government colleges are seeking approval to launch new undergraduate courses in data science, artificial intelligence, and defence studies. The colleges want to attract top students ahead of the 2026-27 academic session. The new courses aim to give students practical skills and real-world project experience. They also plan to collaborate with industry experts. The introduction of these courses is subject to government approval.
Sources
- Amazon Pursues $8 Billion Sale-and-Leaseback of Nvidia Chips
- Amazon Wants to Move $8 Billion in Nvidia Chips Off Its Books. Is AI's Financing Model Starting to Change?
- Amazon plans to move $8B Nvidia chips off its books via leaseback, signaling a shift in AI hardware financing.
- The Bull Case For Innodata (INOD) Could Change Following New Physical AI Motion-Capture Lab Opening
- How Investors May Respond To Innodata (INOD) Opening Robot Data Lab
- Eisman says AI bubble rests on two customers
- Masayoshi Son’s AI ambitions outgrow SoftBank’s balance sheet
- Former Anthropic researcher Coxon to testify at New York City AI hearing, Bloomberg News reports
- OpenAI President Greg Brockman Pulls Planned Second $25 Million Donation to AI Super PAC, Says He Felt 'Terrible': Report
- 10 Free Generative AI Courses to Take Online in 2026
- ECB Warns AI Risks Are Starting to Converge Across Financial Markets
- City govt colleges seek new courses such as data science, AI and defence studies to attract toppers
Comments
Please log in to post a comment.