【AI big orders aren’t just for GPUs—where is the money going?】
Whenever AI compute comes up, everyone only focuses on Nvidia, possibly missing another spending item: CPUs and cloud services that actually make applications run.
On September 24, Akamai announced that it is expanding its partnership with Anthropic, signing a seven-year contract totaling $11.6 billion in commitments to serve CPU workload needs. The announcement also said that this agreement does not change the company’s revenue guidance for 2026.
These two facts should be viewed together: customers are willing to buy compute over the long term, which indicates there is demand; but the total value of a multi-year contract cannot be simply inserted directly into this year’s income statement.
My concern is whether AI spending, after training chips, will keep spreading into inference, networking, and application execution. Suppliers still need to deliver—order totals aren’t the same as profit.
It’s the same in crypto markets. BTC, ETH, and SOL may fluctuate with tech-stock risk appetite, but getting cloud contracts doesn’t mean those tokens receive equivalent revenue—and it certainly doesn’t mean every coin with “AI” in its name will get a share of the orders.
I will first track enterprises that have both customers and delivery, then see whether risk appetite spreads into the crypto market. You can’t skip steps: turning industry news into a reason to buy coins is not enough.
The story can be hot together, but revenue won’t be split evenly.
Image: Billykamenides represents Akamai, Wikimedia Commons, CC BY-SA 4.0; original image not modified.
$BTC $ETH $SOL #AI compute
Click my avatar to view live orders with actual deals
Whenever AI compute comes up, everyone only focuses on Nvidia, possibly missing another spending item: CPUs and cloud services that actually make applications run.
On September 24, Akamai announced that it is expanding its partnership with Anthropic, signing a seven-year contract totaling $11.6 billion in commitments to serve CPU workload needs. The announcement also said that this agreement does not change the company’s revenue guidance for 2026.
These two facts should be viewed together: customers are willing to buy compute over the long term, which indicates there is demand; but the total value of a multi-year contract cannot be simply inserted directly into this year’s income statement.
My concern is whether AI spending, after training chips, will keep spreading into inference, networking, and application execution. Suppliers still need to deliver—order totals aren’t the same as profit.
It’s the same in crypto markets. BTC, ETH, and SOL may fluctuate with tech-stock risk appetite, but getting cloud contracts doesn’t mean those tokens receive equivalent revenue—and it certainly doesn’t mean every coin with “AI” in its name will get a share of the orders.
I will first track enterprises that have both customers and delivery, then see whether risk appetite spreads into the crypto market. You can’t skip steps: turning industry news into a reason to buy coins is not enough.
The story can be hot together, but revenue won’t be split evenly.
Image: Billykamenides represents Akamai, Wikimedia Commons, CC BY-SA 4.0; original image not modified.
$BTC $ETH $SOL #AI compute
Click my avatar to view live orders with actual deals
