• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
Technology
Report

SoftBank Completes $10B OpenAI Investment; Broadcom to Lend Anthropic $42B for TPUs

SoftBank finalized a $10B tranche of its OpenAI commitment (total ~$30B for ~13% stake). Broadcom agreed to lend Anthropic up to $42B to finance TPU infrastructure leases, creating a circular financing structure. The deals underscore AI's massive capital requirements.

TK
RE
FX
4 Sources, 4h ago, first seen 4h ago

TLDR

These transactions reflect the AI arms race's extreme capital intensity—compute, memory, and power are now as critical as models themselves. The deals highlight ecosystem dependencies (TPU vs. GPU), potential valuation bubbles, and risks of technology obsolescence. They reveal new financing models using GPU collateral and residual value insurance, reshaping cloud and AI economics while raising questions about financial sustainability.

Combined views

83K

4 Sources, first seen 4h ago

379 likes37 comments41 saves42 reposts
Featured Source

Combined views

83K

4 Sources, first seen 4h ago

379 likes37 comments41 saves42 reposts

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

4 Sources

@firstadopterOH SNAP. Nvidia wrote a blog post on how Blackwell GPUs are powering GPT-6 Astra Ultrafast with a quote from OpenAI inference lead. Where you at @cerebras ? “NVIDIA’s deep investment in tooling and documentation has enabled us to make our models exceptionally good at programming Blackwell and Rubin GPUs,” said Philippe Tillet, inference lead at OpenAI. “Astra can turn that knowledge into high-performance kernels that make NVIDIA hardware compelling across the full frontier of latency, throughput and cost. With Astra Ultrafast, that means faster model responses as agents write code, use tools and work through complex tasks.” “Our work with NVIDIA is helping us make AI faster and more useful,” said Uday Ruddarraju, chief technology officer of compute at OpenAI." https://blogs.nvidia.com/blog/gpus-openai-gpt-6-astra-ultrafast/?ncid=so-twit-351193
@ReutersEXCLUSIVE: Broadcom to lend Anthropic up to $42 billion to lease its chips, filing says https://reut.rs/46VJqcG https://reut.rs/46VJqcG
@fxevolution5 Things That Caught My Attention Today... 1. This might be one of the clearest examples yet of what people mean when they talk about circular financing in AI. Broadcom has agreed to lend Anthropic up to $42 BILLION to help finance its infrastructure buildout, including a five-year lease for enormous amounts of TPU compute. Anthropic then spends heavily on technology supplied by Broadcom, and could become Broadcom's largest compute customer next year. The debt can even potentially convert into Anthropic equity. 2. Google just released Gemini 4 Argon, and some of the numbers are pretty extraordinary. Google says it scored 77.9% on DeepSWE for long-horizon software engineering, 88.8% on LABBench 2 and has increased maximum output to an enormous 1 million tokens. Internally, Google says Argon has already helped free more than 300 TiB of data-centre memory. Interestingly, Google isn't releasing its full capabilities to everyone immediately, the first access is going to trusted cybersecurity defenders because of what the model can do autonomously. 3. Here's another part of the AI financing story that I think deserves more attention. Nvidia has been exploring ways for lenders and insurers to become more comfortable financing AI infrastructure where GPUs themselves effectively become collateral. The problem is useful life. AI companies would obviously benefit if lenders treated GPUs more like long-lived infrastructure, but lenders have to think about how quickly today's cutting-edge chips could become obsolete or even start breaking. The FT reports insurers are even exploring residual-value insurance to protect lenders if the GPUs backing these loans lose value faster than expected. When you're talking about hundreds of billions in infrastructure, how Wall Street values a GPU three or five years from now suddenly matters a lot. 4. Apparently even the Big Mac is becoming an AI story. Reuters found McDonald's is using a machine-learning pricing engine that analyses millions of transactions and recommends an “optimal price” for individual restaurants, including estimates of local customers' willingness to pay. Reuters found one company-owned Fresno store charging $5.69 for a Big Mac and another just two miles away charging $6.89. Importantly, McDonald's says AI doesn't actually set the final price, franchisees do, and Reuters couldn't establish that the AI caused that particular difference. 5. The US dollar has quietly hit an 18-month high. DXY briefly touched 102 as US growth remains relatively strong and Treasury yields stay elevated, while weakness in Europe has added pressure to the euro. We've spent a lot of time looking at the 10-year and 30-year Treasury recently, but the dollar is another piece of the same puzzle. If this move continues, it's something I'll be watching across commodities, multinational earnings and global financial conditions. Patience. React, don't predict.
@KalshiJUST IN: SoftBank wires final $10B into OpenAI — $64.6B invested for ~13% stake
    • HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI

    4 Sources

    @firstadopterOH SNAP. Nvidia wrote a blog post on how Blackwell GPUs are powering GPT-6 Astra Ultrafast with a quote from OpenAI inference lead. Where you at @cerebras ? “NVIDIA’s deep investment in tooling and documentation has enabled us to make our models exceptionally good at programming Blackwell and Rubin GPUs,” said Philippe Tillet, inference lead at OpenAI. “Astra can turn that knowledge into high-performance kernels that make NVIDIA hardware compelling across the full frontier of latency, throughput and cost. With Astra Ultrafast, that means faster model responses as agents write code, use tools and work through complex tasks.” “Our work with NVIDIA is helping us make AI faster and more useful,” said Uday Ruddarraju, chief technology officer of compute at OpenAI." https://blogs.nvidia.com/blog/gpus-openai-gpt-6-astra-ultrafast/?ncid=so-twit-351193
    @ReutersEXCLUSIVE: Broadcom to lend Anthropic up to $42 billion to lease its chips, filing says https://reut.rs/46VJqcG https://reut.rs/46VJqcG
    @fxevolution5 Things That Caught My Attention Today... 1. This might be one of the clearest examples yet of what people mean when they talk about circular financing in AI. Broadcom has agreed to lend Anthropic up to $42 BILLION to help finance its infrastructure buildout, including a five-year lease for enormous amounts of TPU compute. Anthropic then spends heavily on technology supplied by Broadcom, and could become Broadcom's largest compute customer next year. The debt can even potentially convert into Anthropic equity. 2. Google just released Gemini 4 Argon, and some of the numbers are pretty extraordinary. Google says it scored 77.9% on DeepSWE for long-horizon software engineering, 88.8% on LABBench 2 and has increased maximum output to an enormous 1 million tokens. Internally, Google says Argon has already helped free more than 300 TiB of data-centre memory. Interestingly, Google isn't releasing its full capabilities to everyone immediately, the first access is going to trusted cybersecurity defenders because of what the model can do autonomously. 3. Here's another part of the AI financing story that I think deserves more attention. Nvidia has been exploring ways for lenders and insurers to become more comfortable financing AI infrastructure where GPUs themselves effectively become collateral. The problem is useful life. AI companies would obviously benefit if lenders treated GPUs more like long-lived infrastructure, but lenders have to think about how quickly today's cutting-edge chips could become obsolete or even start breaking. The FT reports insurers are even exploring residual-value insurance to protect lenders if the GPUs backing these loans lose value faster than expected. When you're talking about hundreds of billions in infrastructure, how Wall Street values a GPU three or five years from now suddenly matters a lot. 4. Apparently even the Big Mac is becoming an AI story. Reuters found McDonald's is using a machine-learning pricing engine that analyses millions of transactions and recommends an “optimal price” for individual restaurants, including estimates of local customers' willingness to pay. Reuters found one company-owned Fresno store charging $5.69 for a Big Mac and another just two miles away charging $6.89. Importantly, McDonald's says AI doesn't actually set the final price, franchisees do, and Reuters couldn't establish that the AI caused that particular difference. 5. The US dollar has quietly hit an 18-month high. DXY briefly touched 102 as US growth remains relatively strong and Treasury yields stay elevated, while weakness in Europe has added pressure to the euro. We've spent a lot of time looking at the 10-year and 30-year Treasury recently, but the dollar is another piece of the same puzzle. If this move continues, it's something I'll be watching across commodities, multinational earnings and global financial conditions. Patience. React, don't predict.
    @KalshiJUST IN: SoftBank wires final $10B into OpenAI — $64.6B invested for ~13% stake
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet