Tomer Tunguz 分析 AI 算力短缺：瓶颈已从 GPU 转向电力、数据中心与内存
Hello, Claude? Are You There?
Tomer Tunguz 撰文指出 AI 基础设施短缺已不止是 GPU，还包括电力、数据中心、内存和 CPU，并引用 Sam Altman、Safra Catz、Satya Nadella、Sundar Pichai 与 Intel CEO Lip-Bu Tan 的表态，其中 Lip-Bu Tan 称到 2028 年前都无缓解。
In short : The AI infrastructure shortage isn't chips anymore — it's power, data centers & memory. What happens when demand exceeds supply through 2028?
Feb 2025
"We've been growing a lot and are out of GPUs."
Sam Altman, OpenAI CEO 1
Mar 2025
"We are still waving off customers or scheduling them out into the future. This is a situation that we have not seen in our history."
Safra Catz, Oracle CEO 2
Oct 2025
"You may actually have a bunch of chips sitting in inventory that I can't plug in. I don't have warm shells to plug into."
Satya Nadella, Microsoft CEO 3
Feb 2026
"What keeps us up at night… The top question is definitely around capacity. All constraints — be it power, land, supply chain constraints — how do you ramp up to meet this extraordinary demand?"
Sundar Pichai, Alphabet CEO 4
Feb 2026
"There's no relief as far as I know. No relief until 2028."
Lip-Bu Tan, Intel CEO 5
What happens when your AI doesn’t answer?
Everything is in short supply. It’s no longer just GPUs. It’s power. Data centers. Memory. CPUs.
If there’s no relief for six more quarters, perhaps it’s time to plan for a world where inference isn’t freely available on-demand.
Inference prices, which have been static, will rise. Subsidies will be harder to justify.
Enterprises will need to rationalize workloads, deciding which teams receive state-of-the-art models & which don’t. Not every CRM update requires a trillion-parameter frontier model.
Inference rationing normalizes. Marketing receives this much, sales receives that much, software engineers probably receive a lot more.
Constraint will be the mother of invention. Companies will optimize what they have, adopt open source where they can, and likely move to smaller models for many workloads.
$ Hello, Claude. Are you there?
Waiting until 2028...
Get the next one in your inbox
The 1-minute read that turns tech data into strategic advantage.
Read by 150k+ founders & operators.
GP at Theory Ventures. Former Google PM. Sharing data-driven insights on AI, web3, & venture capital.
来源：Tomer Tunguz 博客（VC 分析） · tomtunguz.com