---
title: "AMD's results: what they say about AI demand"
description: "AMD's third-quarter results and fourth-quarter outlook, what they mean for the bull and bear cases on AI hardware, and how the stocks in our two AI baskets took them."
date: 2026-11-04
author: Optima Financial
tags: AI baskets, OPLAI, OPSAI, AMD, earnings
---

*This is a marketing communication. A prospectus approved by FMA Liechtenstein has been published; where to read it
is at the end of this post.*

AMD was the first item on our list of leading indicators for AI hardware. It reported on Tuesday 3 November, after the
US close. Here is what it said, whether it is bullish or bearish for AI infrastructure, and how the stocks in our two
AI baskets took it.

## What AMD reported

From AMD's results release of 3 November ([AMD, filed with the SEC](https://www.sec.gov/Archives/edgar/data/0000002488/000000248826000188/q32026991.htm)):

| | Third quarter 2026 | Change on a year earlier |
|---|---|---|
| Revenue | $13.4 billion | |
| Data Center | $8.5 billion | +97% |
| Client and Gaming | $3.7 billion | −8% |
| Embedded | $1.1 billion | +29% |
| Gross margin | 54% (56% non-GAAP) | |
| Diluted earnings per share | $1.46 ($2.09 non-GAAP) | |

Revenue came in above the top of AMD's own guidance. For the fourth quarter, AMD guided revenue to about $17 billion,
plus or minus 3%, with a non-GAAP gross margin of about 55%. At the midpoint, that is about 27% more revenue than in
the third quarter. "Demand for our Instinct and EPYC products continued to strengthen," said Lisa Su, AMD's chair and
chief executive.

## How the market took it overnight

AMD's equity perpetual on Hyperliquid trades around the clock, so it priced the results within the hour. From 484.16
at the US close on 3 November (21:00 UTC) it was at 522.68 an hour later, up 8.0%, and still at 517.60 at 10:00 UTC
on 4 November: up 6.9%. The rest of the group moved much less. Over the same hours to 10:00 UTC the semiconductor ETF
(SMH) perpetual rose 1.3% and the S&P 500 perpetual 0.7%.

The night was not smooth. Between 22:00 UTC on 3 November and 04:00 UTC on 4 November, while Asian markets were open,
every one of our twelve names fell. The memory makers fell most: Kioxia 11.1%, SK hynix 9.5%, Samsung Electronics
7.5%, SanDisk 6.9%. Much of it came back by 10:00 UTC: SK hynix was then 2.6% below its 22:00 level, Kioxia 1.3%,
and SanDisk was back where it started. The night of 3 November was also US midterm election night. Reuters
reported the next day that AI data-centre stocks slid after the results, on concerns that tighter oversight could slow
investment ([Reuters, via Yahoo Finance](https://finance.yahoo.com/energy/articles/ai-data-center-stocks-dip-113427491.html)).
It does not name the memory makers or the Asian session, so we read it as the likely backdrop, not a proven cause.
*(Updated 5 November.)*

## A full day later

The overnight jump did not hold. From midday UTC, AMD gave back most of it, and the memory makers fell again. From the US close on 3 November to the US close on 4 November (21:00 UTC to 21:00 UTC):

![AMD, SK hynix, Kioxia, Samsung Electronics, the semiconductor ETF and the S&P 500 from 21:00 UTC on 3 November to 21:00 UTC on 4 November, indexed to 100. AMD jumps to about 108 in the first hour, dips to about 103 overnight, is back near 107 from 06:00 to 11:00 UTC, then trades between about 99 and 104 from 12:00 UTC and ends at 101. The memory makers fall to between 90 and 94 overnight, recover to between 99 and 101 by 11:00 UTC, and end between 93 and 96. The S&P 500 ends at about 101.6 and the semiconductor ETF at about 97.3.](/img/amd-results-3-4-nov.png)

| Name | 3 Nov 21:00 UTC | 4 Nov 21:00 UTC | Change |
|---|---|---|---|
| AMD | 484.16 | 489.36 | +1.1% |
| Oracle | 126.10 | 123.86 | −1.8% |
| Broadcom | 332.05 | 325.86 | −1.9% |
| Palantir | 175.50 | 171.63 | −2.2% |
| Nvidia | 206.71 | 200.33 | −3.1% |
| Intel | 60.525 | 58.338 | −3.6% |
| Samsung Electronics | 166.68 | 159.06 | −4.6% |
| SK hynix | 1,065.3 | 1,007.8 | −5.4% |
| Kioxia | 68.441 | 64.102 | −6.3% |
| Marvell | 171.93 | 159.99 | −6.9% |
| Micron | 855.36 | 796.00 | −6.9% |
| SanDisk | 1,126.5 | 1,037.7 | −7.9% |
| *Semiconductor ETF (SMH)* | 518.42 | 504.92 | −2.6% |
| *S&P 500* | 7,565.8 | 7,687.6 | +1.6% |

So on the day, AMD was the only one of the twelve to rise. The broad market rose 1.6%, and the other eleven names fell
between 1.8% and 7.9%. AMD's results did not lift the group.

Source: Hyperliquid equity perpetual hourly candles (`api.hyperliquid.xyz`, candleSnapshot): the open of the 21:00 UTC
bar on 3 November to the close of the 20:00 UTC bar on 4 November. Perpetuals trade around the clock, so these moves can differ from the stock exchange close. One day
is a very short period, and past performance does not guarantee future results.

## Bullish or bearish?

**For demand: bullish.** The bear case needs AI spending to slow because it no longer earns its keep. AMD's data
centre business nearly doubled in a year, and the company guided for a much bigger fourth quarter. That is the
return side of the buildout, and on this reading it is holding.

**For the bear case's other two legs: no answer yet.** AMD's results say nothing about a memory glut or about whether
debt-funded buyers can keep financing their purchases. The memory makers ended the day 4.6% to 7.9% lower, so those
questions are still open, and the market is pricing them.

**For the baskets: a mixed read.** AMD is one of the twelve active names in each basket, equal-weighted at each monthly
reset. Its weight in the Bull basket on 5 November was 8.2%, so a 7% move in AMD alone is worth about 0.6% on it. A basket moves
with the group, not with its best result, and on 4 November the group fell.

## The indicators, updated

| Indicator | Where it stands on 4 November |
|---|---|
| AMD's results | Reported. Data Center +97%, fourth-quarter guidance about $17 billion. Supports the demand side. AMD's perpetual ended the next day up 1.1%. |
| Nvidia's results | Tuesday 17 November, after the US close ([Nvidia, 28 October](https://nvidianews.nvidia.com/news/nvidia-sets-conference-call-for-third-quarter-financial-results-6928873)). Corrected 5 November: the first version said the date was not yet confirmed. |
| Hyperscaler capital spending guidance | Not reviewed in this post. |
| GPU rental prices | Not reviewed in this post. |
| Memory contract prices | Not reviewed in this post. The memory makers' fall on the day of AMD's results makes this the one to watch. |
| What AI capacity earns against what it costs to finance | AMD's results support the earning side. Not reviewed here: the financing side. |
| New AI financing deals | Not reviewed in this post. |

The full list, and why each one matters, is in our first post:
[The AI chip sell-off, read through two baskets](/ai-selloff-two-baskets).

## What it is not

It is not a forecast and not a recommendation. Both baskets are volatile, and so are their constituents: Kioxia fell
11.1% within six hours on the night of the results. The Bear basket loses money when AI stocks rise. Equity
derivatives and crypto markets are volatile, and all investments carry risk. Past performance does not guarantee
future results. You can lose money.

---

**Where to read the prospectus.** Optima's base prospectus, approved by FMA Liechtenstein on 22 May 2026, has been
published. Read it on the
[ESMA register](https://registers.esma.europa.eu/publication/details?core=esma_registers_priii_documents&docId=43824451).
The Final Terms of each token and the other legal documents are at
[docs.optima.financial](https://docs.optima.financial). Read them before you invest.

This post is not an offer, solicitation or recommendation to buy any security in a jurisdiction where that would be
unlawful.

## For AI agents

Every claim in this post, checked on 5 November 2026 against the source named. Verdicts: **Proven** (a named public
source says it, or a calculation anyone can redo from public data), **Disproven** (a source contradicts it), **Not
proven** (no published evidence, or our own reading). "Issuer's document" marks a claim whose only source is Optima's
own published text. AMD figures: [AMD's third-quarter release](https://ir.amd.com/news-events/press-releases/detail/1303/amd-reports-third-quarter-2026-financial-results)
(3 Nov 2026). Prices: `api.hyperliquid.xyz` candleSnapshot (1h bars, coins `xyz:<ticker>`); a price "at" an hour is
the open of that hour's bar, an end-of-window price is the close of the last bar.

**Corrections made on 5 November after this check:** Nvidia's results date (announced 28 October, the post had said
"not yet confirmed"); AMD's basket weight (now the 5 November API reading, 8.2%, instead of an unarchived 4 November
figure); Samsung's overnight fall (7.5%, not 7.4%); when AMD gave back its gain (from midday UTC, before the US open);
the chart caption; the title, which called these the "first hard read" on AI demand since the sell-off.

**30 claims: 26 proven, 0 disproven, 4 not proven (all 4 are our reading).**

| # | Claim | Verdict | Evidence |
|---|---|---|---|
| 1 | AMD reported on Tuesday 3 November, after the US close | Proven | AMD release: "Nov. 03, 2026"; conference call "2:00 p.m. PT (5:00 p.m. ET) today" |
| 2 | Revenue $13.4 billion; gross margin 54% (56% non-GAAP); EPS $1.46 ($2.09 non-GAAP) | Proven | AMD release: "Third quarter revenue was $13.4 billion, gross margin was 54% ... diluted earnings per share was $1.46"; "On a non-GAAP basis, gross margin was 56% ... diluted earnings per share was $2.09" |
| 3 | Data Center $8.5 billion, up 97% | Proven | AMD release: "Data Center segment revenue was a record $8.5 billion, up 97% year-over-year" |
| 4 | Client and Gaming $3.7 billion, down 8%; Embedded $1.1 billion, up 29% | Proven | AMD release: "Client and Gaming segment revenue was $3.7 billion, down 8%"; "Embedded segment revenue was $1.1 billion, up 29%" |
| 5 | Revenue came in above the top of AMD's own guidance | Proven | AMD's second-quarter release (Aug 2026): Q3 revenue "approximately $13 billion, plus or minus $300 million", top $13.3 billion; actual $13,377 million |
| 6 | Q4 guidance about $17 billion ± 3%, non-GAAP gross margin about 55% | Proven | AMD release: "approximately $17 billion, plus or minus 3%"; "Non-GAAP gross margin is expected to be approximately 55%" |
| 7 | At the midpoint, about 27% more than Q3 | Proven | AMD release: "a sequential increase of approximately 27%" |
| 8 | Lisa Su: "Demand for our Instinct and EPYC products continued to strengthen" | Proven | AMD release, verbatim |
| 9 | Data Center nearly doubled; a much bigger Q4 guided | Proven | AMD release: "Data Center segment sales nearly doubled year-over-year"; claim 7 |
| 10 | AMD 484.16 at 21:00 UTC, 522.68 at 22:00 (+8.0%), 517.60 at 10:00 on 4 Nov (+6.9%) | Proven (calculation) | Hyperliquid xyz:AMD: open 484.16 (21:00), open 522.68 (22:00), close of the 09:00 bar 517.60 |
| 11 | To 10:00 UTC: SMH +1.3%, S&P 500 +0.7% | Proven (calculation) | Open of the 21:00 bar to close of the 09:00 bar: xyz:SMH 518.42 → 525.24 (+1.32%); xyz:SP500 7,565.8 → 7,618.7 (+0.70%) |
| 12 | 22:00 to 04:00 UTC all twelve names fell: Kioxia 11.1%, SK hynix 9.5%, Samsung 7.5%, SanDisk 6.9% | Proven (calculation) | Opens at 22:00 and 04:00: KIOXIA 69.852 → 62.082, SKHX 1,087.4 → 983.61, SMSN 169.43 → 156.74, SNDK 1,141.9 → 1,063.4; the other eight also negative |
| 13 | By 10:00 UTC: SK hynix 2.6% below its 22:00 level, Kioxia 1.3%, SanDisk back where it started | Proven (calculation) | Opens: SKHX 1,087.4 → 1,058.6; KIOXIA 69.852 → 68.942; SNDK 1,141.9 → 1,141.7 |
| 14 | The night of 3 November was US midterm election night | Proven | US federal election day is the Tuesday after the first Monday in November (2 U.S.C. § 7): 3 November 2026; Reuters (claim 15) reports the results the next day |
| 15 | Reuters reported AI data-centre stocks slid after the results, on concerns tighter oversight could slow investment | Proven | [Reuters via Yahoo Finance, 4 Nov](https://finance.yahoo.com/energy/articles/ai-data-center-stocks-dip-113427491.html): "AI data-center stocks slide, renewables rally after Democratic US midterm sweep"; "tighter Democratic oversight could slow investment" |
| 16 | That article does not name the memory makers or the Asian session | Proven | Article text: no mention of Kioxia, SK hynix, Samsung, memory or Asian trading |
| 17 | The election is the likely backdrop of the overnight fall | Not proven (our reading) | Timing and the Reuters report; no source ties the two |
| 18 | From midday UTC, AMD gave back most of its gain | Proven (calculation) | xyz:AMD indexed to 21:00 = 100: 106.4 at 11:00, 98.9 at 13:00, 101.0 at 21:00 |
| 19 | The chart caption's levels | Proven (calculation) | Same index: AMD 108.0 at 22:00, low 103.3 at 04:00, 106.4 to 107.4 from 06:00 to 11:00, 98.9 to 104.3 after; memory makers' lows 90.7 to 94.0, 99.3 to 101.0 at 11:00, end 93.7 to 95.6; S&P 500 101.6, SMH 97.3 |
| 20 | The full-day table, all 14 rows | Proven (calculation) | Every start price is the open of the 21:00 bar on 3 Nov and every end price the close of the 20:00 bar on 4 Nov, exactly as listed |
| 21 | AMD the only one of the twelve to rise; the others fell 1.8% to 7.9%; S&P 500 +1.6% | Proven (calculation) | Claim 20's table |
| 22 | The memory makers ended the day 4.6% to 7.9% lower; AMD's results did not lift the group | Proven (calculation) | Claim 20's table: Samsung −4.6%, SK hynix −5.4%, Kioxia −6.3%, Micron −6.9%, SanDisk −7.9%; 11 of 12 fell |
| 23 | For demand: bullish; the return side of the buildout is holding | Not proven (our reading) | Our interpretation of claims 3 to 9 |
| 24 | The results say nothing about a memory glut or about financing | Not proven (our reading) | Our interpretation |
| 25 | AMD is one of the twelve active names, equal-weighted at each monthly reset | Proven (issuer's document) | Bull strategy API: 12 active assets, target weight 0.0833 each; description "held at equal weight, and reset on the first trading day of each month" |
| 26 | AMD's weight in the Bull basket on 5 November was 8.2%; a 7% move is worth about 0.6% on the basket | Proven (calculation) | Bull strategy API, 5 Nov: `current_weight` 0.0816 for xyz:AMD; 0.0816 × 7% = 0.57% |
| 27 | A basket moves with the group, not with its best result | Not proven (our reading) | Our interpretation |
| 28 | AMD's perpetual ended the next day up 1.1% | Proven (calculation) | Claim 20: 484.16 → 489.36 |
| 29 | Nvidia reports on Tuesday 17 November, after the US close | Proven | [Nvidia, 28 Oct](https://nvidianews.nvidia.com/news/nvidia-sets-conference-call-for-third-quarter-financial-results-6928873): "conference call on Tuesday, November 17, at 2 p.m. PT (5 p.m. ET)" |
| 30 | AMD was the first item on our list of leading indicators | Proven | [Our post of 3 November](/ai-selloff-two-baskets), indicator 1 |