Thesis · Hyperscalers, Cloud & Compute Services

Will Microsoft, Amazon, or Alphabet reveal that one customer makes up 10% or more of their revenue or future contracted business?

A handful of richly funded AI labs may already loom so large in cloud sales that one of the big three cloud providers has to name them in official filings before the end of 2027.

6 dated tests · 0 resolved · next on

The tests More theses in Hyperscalers, Cloud & Compute Services

Who is on each side

The two sides.

Each side in DeepStack’s words, then the voices who argue it, attributed and paraphrased, dated where the date is known. A voice links to its page on DeepStack where one exists.

The case for yes

A single AI lab crosses the 10% disclosure line

One of the three major cloud providers states in an official filing that a single customer makes up 10% or more of total company revenue or of signed future business. This would confirm that AI cloud growth leans heavily on a small set of labs rather than on a wide base of paying enterprises.

The case for no

No single customer crosses the 10% line

None of the three cloud providers names a customer crossing the 10% mark in a filing through 2027. On this view the absence of such a disclosure is itself the evidence: cloud revenue growth across the big three continues to be reported in aggregate terms, with no filing attributing 10% or more of revenue or signed future business to any one customer.

1 voice

A single AI lab crosses the 10% disclosure line

  1. Ed Elson

    Circular AI revenue dependent on subsidised buyers

6 voices

No single customer crosses the 10% line

  1. Hf Best Ideas

    The fund's argument is that heavy build-out depresses free cash flow during construction, and the test is not the cash flow line but whether the finished asset becomes infrastructure competitors must pay to use; Cloud growth, margin expansion and…

  2. Invest in Assets

    Capex is rewarded only when returns are visible

  3. Ren

    Microsoft holds capex guidance, Amazon grows AWS 36.7% YoY while expanding capex further. These are the two largest cloud providers maintaining their capex expansion precisely when bears argued growth and capex expansion could not coexist. The…

  4. DeepValue Capital

    Building or buying chips and data centers adds capacity that is immediately absorbed by AI agent and coding demand; because customers are still queuing, prices can be raised while unit costs fall through optimization, a higher MaaS mix and internal…

  5. Michael Spencer

    Amazon invests capex in advance of customer monetization, laying out cash for land, power, buildings, chips, and networking gear before billing customers. These assets have long useful lives (30+ years for datacenters), creating a cycle where high…

  6. Harris Associates Oakmark

    Alphabet's multi-year AI spending is producing results visible in rapid Google Cloud Services growth and operating expense reduction. Despite the spending raising depreciation from 3.7% to 5.3% of revenue, operating margins expanded 4 percentage…

What would settle it

The dated tests.

Each test names the public source that settles it, the rule and the day. The Call’s number, the readers’ median and the machines’ numbers stand beside it, as on the Docket.

  1. CoreWeave, purchases of property and equipment as a share of revenue, quarter ended September 30, 2026: whether it comes in at or above 225%.

    On the Docket Pendingth030-t01Open question

    Specification pending

    The machinesClaude 40% · DeepSeek 87%

    Claude: 40%DeepSeek: 87%

    Your number

  2. CoreWeave, purchases of property and equipment as a share of revenue, quarter ended September 30, 2026: whether it comes in at or above 248%.

    On the Docket Pendingth030-t02Open question

    Specification pending

    The machinesClaude 33% · DeepSeek 54%

    Claude: 33%DeepSeek: 54%

    Your number

  3. CoreWeave, purchases of property and equipment as a share of revenue, quarter ended September 30, 2026: whether it comes in at or above 270%.

    On the Docket Pendingth030-t03Open question

    Specification pending

    The machinesClaude 25% · DeepSeek 40%

    Claude: 25%DeepSeek: 40%

    Your number

  4. Oracle, purchases of property and equipment as a share of revenue, quarter ended November 30, 2026: whether it comes in at or above 130%.

    On the Docket Pendingth030-t04Open question

    Specification pending

    The machinesClaude 33% · DeepSeek 52%

    Claude: 33%DeepSeek: 52%

    Your number

  5. Oracle, purchases of property and equipment as a share of revenue, quarter ended November 30, 2026: whether it comes in at or above 148%.

    On the Docket Pendingth030-t05Open question

    Specification pending

    The machinesClaude 22% · DeepSeek 30%

    Claude: 22%DeepSeek: 30%

    Your number

  6. Oracle, purchases of property and equipment as a share of revenue, quarter ended November 30, 2026: whether it comes in at or above 165%.

    On the Docket Pendingth030-t06Open question

    Specification pending

    The machinesClaude 6% · DeepSeek 3%

    Claude: 6%DeepSeek: 3%

    Your number

In The Buildout

The stories that take up this thesis, with their own Calls and tests.

A thesis page states one question and both sides of it; DeepStack takes no side in words, and its view is the number it gives each test before the date. Analysis and opinion, not investment advice. Voices are paraphrased from public writing and attributed by name; DeepStack shows no score, ranking or accuracy for any person. Pre-specified means the words were approved on or before the test’s date. Specified after the date means they were written afterwards: such a test is counted and shown, and never rated. An open question has no words yet. The Call’s number is the chance, from 0 to 100, that the Call gives the yes answer to a test’s question, recorded before the test’s date; it is DeepStack’s own reading, not a market price and not advice. The readers’ mark on the same line is the readers’ median, the middle of the numbers signed-in readers gave the test, shown from 25 readers. The machines: Each model read this page and nothing else. The number that counts is the latest one each model gave before the test closed.