Thesis · Semiconductors & Compute Silicon

Will Broadcom's AI semiconductor revenue for fiscal 2027 (ending ~Oct/Nov 2027), as reported in its quarterly and annual results, exceed $60 billion?

Hyperscaler custom accelerator programs scale fast enough that the leading custom-silicon designer's AI revenue triples from roughly $20B in FY25 to above $60B by FY27, evidencing real share capture from merchant GPUs.

6 dated tests · 0 resolved · next on

The tests More theses in Semiconductors & Compute Silicon

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

Custom silicon takes share from merchant GPUs

Hyperscalers fund multi-year XPU programs because a few billion of ASIC NRE is trivial against hundreds of billions of data-center capex and per-chip cost runs roughly a third of a merchant GPU. Custom accelerator unit volumes approach or exceed GPU volumes and the fixed-fee plus per-chip economics accrue to the two firms capable of complex AI ASIC design.

The case for no

The case against it

Uncertainty about the future mix of ordinary-latency and extreme-interactivity demand, combined with software able to carve a speed tier out of an existing accelerator pool, so buyers reallocating GPU capacity by configuration instead of purchasing dedicated dataflow silicon, compressing the addressable market for specialist chips without eliminating its top end. Nvidia's desire to maintain architectural control as custom AI systems proliferate, so custom systems remain locked into Nvidia infrastructure even as customers demand semi-custom alternatives. It would show first if broadcom AI revenue growth decelerates to under 50% year over year in FY27 with guidance cut.

3 voices

Custom silicon takes share from merchant GPUs

  1. Hf Best Ideas

    Baron points to a Google agreement through 2031, larger planned compute at Anthropic, OpenAI's first inference chip taped out in nine months toward a 10 gigawatt build, Meta's continued investment posture and a multi-year Apple agreement, alongside…

    Source

  2. Market Sentiment

    As buyers stop paying for frontier models on every task, competition turns to lowest cost per token; inference is repetitive enough to hard-wire, and at hyperscaler volume the $500M+ design cost is amortized, so the roughly half of inference cost…

    Source

  3. Invest in Assets

    Broadcom monetizes the AI race itself rather than any single winner: every lab and hyperscaler that keeps competing must buy XPUs and networking, so customer-level outcomes matter less than the continuation of the race.

    Source

2 voices

The case against it

  1. Gaetano

    By investing in and integrating component suppliers like Marvell across custom silicon, scale-up networking, and photonic fabric, Nvidia ensures downstream AI infrastructure remains dependent on Nvidia rails regardless of customer preferences for…

    Source

  2. Dylan Patel

    By combining purpose-built hardware accelerators with flexible software compilation and task scheduling frameworks, a vendor can achieve the performance benefits of custom solutions while preserving the programmability and portability advantages of…

    Source

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. Marvell, revenue growth from a year earlier, quarter ended October 31, 2026: whether it comes in at or above 46%.

    On the Docket Pendingth016-t04Open question

    Specification pending

    The machinesClaude 55% · DeepSeek 55%

    Claude: 55%DeepSeek: 55%

    Your number

  2. Marvell, revenue growth from a year earlier, quarter ended October 31, 2026: whether it comes in at or above 54%.

    On the Docket Pendingth016-t05Open question

    Specification pending

    The machinesClaude 33% · DeepSeek 28%

    Claude: 33%DeepSeek: 28%

    Your number

  3. Marvell, revenue growth from a year earlier, quarter ended October 31, 2026: whether it comes in at or above 62%.

    On the Docket Pendingth016-t06Open question

    Specification pending

    The machinesClaude 14% · DeepSeek 7% · GPT 24%

    Claude: 14%DeepSeek: 7%GPT: 24%

    Your number

  4. Broadcom, revenue growth from a year earlier, quarter ended November 1, 2026: whether it comes in at or above 82%.

    On the Docket Pendingth016-t01Pre-specified

    Question
    Did the revenue of Broadcom for the quarter ended November 1, 2026 grow at or above 82.0% from the quarter ended November 2, 2025?
    Source
    SEC EDGAR, XBRL company facts of Broadcom (Forms 10-Q and 10-K)
    Rule
    Compare the revenue that Broadcom reports for the quarter ended November 1, 2026 with the quarter ended November 2, 2025, in its 10-Q or 10-K filing on SEC EDGAR. The answer is yes if growth is at or above 82.0%; no if it is below 82.0%.
    If the source is late, revised or silent
    If Broadcom has not filed by January 7, 2027, or a filing restates a period the rule reads, the test is inconclusive, and the Result says what was read and what was missing.
    Base rate
    Broadcom's revenue growth from a year earlier was 85.5% for the quarter ended August 2, 2026, by its filing on SEC EDGAR.
    Outside view
    The year-earlier quarter is not on the same concept in the filings, so the rungs sit on the last print alone.
    What would change the number
    The results-calendar consensus for the quarter ended November 1, 2026 implies 93.7%, and the middle rung is centred there.

    The Call, before the dateExpects yes. Gives the yes answer 78%.

    78%

    0 · no100 · yes

    The machinesClaude 18% · DeepSeek 10% · GPT 46%

    Claude: 18%DeepSeek: 10%GPT: 46%
    The ladder 3 rungs on one reading, Dec 9, 2026: each asks whether it lands above its threshold, so the Call’s numbers down the rungs sketch where it expects the reading to land.

    1. Above 82%Call 78% · Claude 18% · DeepSeek 10% · GPT 46%

      78%

    2. Above 93%Call 52% · Claude 4% · DeepSeek 5% · GPT 7%

      52%

    3. Above 103%Call 27% · Claude 5% · DeepSeek 8% · GPT 7%

      27%

    0 · no50100 · yes

    Your number

  5. Broadcom, revenue growth from a year earlier, quarter ended November 1, 2026: whether it comes in at or above 93%.

    On the Docket Pendingth016-t02Pre-specified

    Question
    Did the revenue of Broadcom for the quarter ended November 1, 2026 grow at or above 93.0% from the quarter ended November 2, 2025?
    Source
    SEC EDGAR, XBRL company facts of Broadcom (Forms 10-Q and 10-K)
    Rule
    Compare the revenue that Broadcom reports for the quarter ended November 1, 2026 with the quarter ended November 2, 2025, in its 10-Q or 10-K filing on SEC EDGAR. The answer is yes if growth is at or above 93.0%; no if it is below 93.0%.
    If the source is late, revised or silent
    If Broadcom has not filed by January 7, 2027, or a filing restates a period the rule reads, the test is inconclusive, and the Result says what was read and what was missing.
    Base rate
    Broadcom's revenue growth from a year earlier was 85.5% for the quarter ended August 2, 2026, by its filing on SEC EDGAR.
    Outside view
    The year-earlier quarter is not on the same concept in the filings, so the rungs sit on the last print alone.
    What would change the number
    The results-calendar consensus for the quarter ended November 1, 2026 implies 93.7%, and the middle rung is centred there.

    The Call, before the dateEither answer fits the Call. Gives the yes answer 52%.

    52%

    0 · no100 · yes

    The machinesClaude 4% · DeepSeek 5% · GPT 7%

    Claude: 4%DeepSeek: 5%GPT: 7%

    Your number

  6. Broadcom, revenue growth from a year earlier, quarter ended November 1, 2026: whether it comes in at or above 103%.

    On the Docket Pendingth016-t03Pre-specified

    Question
    Did the revenue of Broadcom for the quarter ended November 1, 2026 grow at or above 103.0% from the quarter ended November 2, 2025?
    Source
    SEC EDGAR, XBRL company facts of Broadcom (Forms 10-Q and 10-K)
    Rule
    Compare the revenue that Broadcom reports for the quarter ended November 1, 2026 with the quarter ended November 2, 2025, in its 10-Q or 10-K filing on SEC EDGAR. The answer is yes if growth is at or above 103.0%; no if it is below 103.0%.
    If the source is late, revised or silent
    If Broadcom has not filed by January 7, 2027, or a filing restates a period the rule reads, the test is inconclusive, and the Result says what was read and what was missing.
    Base rate
    Broadcom's revenue growth from a year earlier was 85.5% for the quarter ended August 2, 2026, by its filing on SEC EDGAR.
    Outside view
    The year-earlier quarter is not on the same concept in the filings, so the rungs sit on the last print alone.
    What would change the number
    The results-calendar consensus for the quarter ended November 1, 2026 implies 93.7%, and the middle rung is centred there.

    The Call, before the dateExpects no. Gives the yes answer 27%.

    27%

    0 · no100 · yes

    The machinesClaude 5% · DeepSeek 8% · GPT 7%

    Claude: 5%DeepSeek: 8%GPT: 7%

    Your number

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.