
Initiating at Buy, $457 target: the price already pays for most of the AI ramp, and we see 13% more
We initiate at Buy with a $457 target, 13.1% total return and low conviction. AMD designs server and PC processors, Instinct AI accelerators and adaptive chips; Data Center is 81% of FY27E revenue. The $404 price implies FY27E Data Center AI growth of 148%; we model 170%.
- Data Center AI grows 170% to $36.6bn in FY27E. Helios racks ramp at five lead customers; at the Street’s +137% the value is $394, at ours $425. Debate 1.
- Server CPU doubles to $40.6bn on Venice. Guidance is about 2x; each extra 10pp of growth is worth $14 a share.
- Operating margin hits 36% in FY27E; mix and warrants cost $20. Gross margin 54.5%, 1.1pp under the Street. Debates 2 and 3.
What the market is missing. Guidance of 80%+ growth implies FY27E revenue of $93.9bn or more; consensus is $90.8bn, we are $95.3bn (+5.0%); EPS $17.43 vs $16.27 (+7.2%).
Catalysts. NVIDIA results 17 Nov 2026; AMD Q4’26 on 2 Feb 2027 ($16.96bn vs $16.31bn consensus); Q1’27 Helios volume in May.
Valuation and target. 24.4x FY27E EPS (peer PEG 0.22x times 109% growth) gives $425 base. Bear $276, bull $510 at 25/55/20% weights is $405, or $457 rolled one year. DCF: $223.
Risks. US data-center policy −$74, Arm CPUs and NVIDIA Vera −$41, rack mix −$18 a share.
| Year end Dec, US$bn | FY24A | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|---|
| Revenue | 25.8 | 34.6 | 52.1 | 95.3 | 121.3 |
| Growth y/y | 13.7% | 34.3% | 50.5% | 82.9% | 27.3% |
| Non-GAAP op. income | 6.1 | 7.8 | 15.1 | 34.0 | 45.1 |
| Margin | 23.8% | 22.4% | 29.0% | 35.6% | 37.2% |
| Non-GAAP net income | 5.4 | 6.8 | 13.4 | 29.6 | 38.9 |
| Non-GAAP EPS ($) | 3.31 | 4.17 | 8.04 | 17.43 | 22.12 |
| Growth y/y | – | 26.0% | 92.9% | 116.7% | 26.9% |
| Free cash flow | 2.4 | 6.7 | 4.7 | 19.6 | 32.2 |
| At $404.40 | FY24A | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|---|
| P/E | 122.2x | 97.0x | 50.3x | 23.2x | 18.3x |
| EV / non-GAAP EBIT | 108.6x | 85.8x | 44.0x | 19.6x | 14.8x |
| P/B | 11.5x | 10.5x | 9.2x | 6.8x | 5.1x |
| FCF yield | 0.4% | 1.0% | 0.7% | 2.9% | 4.8% |
| Dividend yield | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% |
| ROE (GAAP) | 2.9% | 7.2% | 14.8% | 29.2% | 28.9% |
| Net debt / EBITDA | net cash | net cash | net cash | net cash | net cash |
| P/E at our $457 target | 138.1x | 109.7x | 56.8x | 26.2x | 20.7x |
Source: Company filings (A); Penmark Research model (E); Nasdaq close, 5 Nov 2026. Non-GAAP excludes stock compensation and acquisition amortisation. Shaded = estimates. Fuller ratios in Exhibits 8-9.
Base value is $425 a share today and $480 in 12 months, with a bull-to-bear range of +42% to −23%
Source: Nasdaq closes via Yahoo Finance; Penmark Research model. Case values at 5 Nov 2026 rolled forward one year at the 13.0% cost of equity; target is the probability-weighted value rolled forward.
Server +110%, DC AI +220%; GM 55.5%. FY27E EPS $20.91; $510 today.
Server +100%, DC AI +170%; GM 54.5%. FY27E EPS $17.43; $425 today.
Server +70%, DC AI +90%; GM 53.0%. FY27E EPS $11.31; $276 today.
| Income, US$bn | FY24A | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|---|
| Revenue | 25.8 | 34.6 | 52.1 | 95.3 | 121.3 |
| Gross profit | 13.8 | 18.2 | 29.0 | 52.0 | 66.7 |
| Operating costs | 7.6 | 10.4 | 13.8 | 18.0 | 21.6 |
| EBIT | 6.1 | 7.8 | 15.1 | 34.0 | 45.1 |
| Stock comp. | 1.4 | 1.6 | 2.1 | 2.9 | 3.3 |
| Net inc., adj. | 5.4 | 6.8 | 13.4 | 29.6 | 38.9 |
| Net inc., GAAP | 1.6 | 4.3 | 10.0 | 25.4 | 34.7 |
| D&A | 3.1 | 3.0 | 3.1 | 3.2 | 3.8 |
| Cash flow, US$bn | FY24A | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|---|
| Op. cash flow | 3.0 | 7.7 | 7.2 | 23.9 | 37.0 |
| Capex | (0.6) | (1.0) | (2.5) | (4.3) | (4.9) |
| Free cash flow | 2.4 | 6.7 | 4.7 | 19.6 | 32.2 |
| FCF margin | 9.3% | 19.4% | 8.9% | 20.5% | 26.5% |
| Capex / sales | 2.5% | 2.8% | 4.9% | 4.5% | 4.0% |
| Stock comp. | 1.4 | 1.6 | 2.1 | 2.9 | 3.3 |
| SBC / sales | 5.5% | 4.7% | 3.9% | 3.0% | 2.7% |
| FCF / share ($) | 1.47 | 4.12 | 2.80 | 11.51 | 18.27 |
| Balance, US$bn | FY24A | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|---|
| Cash, investments | 5.1 | 10.6 | 15.7 | 35.3 | 67.5 |
| Receivables | 6.2 | 6.3 | 12.5 | 18.4 | 22.4 |
| Inventories | 5.7 | 7.9 | 14.8 | 20.9 | 23.6 |
| Goodwill, intang. | 43.8 | 41.8 | 40.0 | 38.0 | 36.2 |
| Total assets | 69.2 | 76.9 | 102.4 | 136.9 | 178.2 |
| Debt | 1.7 | 3.2 | 7.1 | 7.1 | 7.1 |
| Net cash | 3.4 | 7.3 | 8.7 | 28.2 | 60.4 |
| Equity | 57.6 | 63.0 | 72.9 | 101.1 | 139.0 |
| Per share, KPIs | FY24A | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|---|
| Shares (m) | 1,637 | 1,636 | 1,660 | 1,700 | 1,761 |
| EPS, adj. ($) | 3.31 | 4.17 | 8.04 | 17.43 | 22.12 |
| EPS, GAAP ($) | 1.00 | 2.65 | 6.04 | 14.93 | 19.69 |
| DPS ($) | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 |
| BVPS ($) | 35.17 | 38.51 | 43.88 | 59.46 | 78.95 |
| Data Center, $bn | 12.6 | 16.6 | 33.9 | 77.2 | 102.0 |
| DC % of sales | 48.8% | 48.0% | 64.9% | 81.0% | 84.1% |
| Revenue growth | 13.7% | 34.3% | 50.5% | 82.9% | 27.3% |
Source: Company filings (A); Penmark Research model (E). Shaded columns are estimates (E). Non-GAAP excludes stock compensation and acquisition amortisation. Net cash = cash and short-term investments less debt. FY26E-FY28E diluted shares include customer warrants as they vest.
Margins widen to 37% EBIT and ROE to 29% as revenue nearly doubles; we sit 3% to 7% above consensus EPS through FY27E
| Growth, margins (%) | FY24A | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|---|
| Growth y/y | |||||
| Sales | 13.7% | 34.3% | 50.5% | 82.9% | 27.3% |
| Non-GAAP EBIT | – | 26.6% | 94.9% | 124.3% | 32.9% |
| Non-GAAP EPS | – | 26.0% | 92.9% | 116.7% | 26.9% |
| Free cash flow | – | 180% | −31% | 320.8% | 64.5% |
| Margins | |||||
| Gross margin | 53.4% | 52.4% | 55.6% | 54.5% | 55.0% |
| EBIT margin | 23.8% | 22.4% | 29.0% | 35.6% | 37.2% |
| Net margin | 21.0% | 19.7% | 25.6% | 31.1% | 32.1% |
| FCF margin | 9.3% | 19.4% | 8.9% | 20.5% | 26.5% |
| Capex / sales | 2.5% | 2.8% | 4.9% | 4.5% | 4.0% |
| Ratios, valuation | FY24A | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|---|
| Returns and leverage | |||||
| ROE (GAAP) | 2.9% | 7.2% | 14.8% | 29.2% | 28.9% |
| ROA (GAAP) | 2.4% | 5.9% | 11.2% | 21.2% | 22.0% |
| ROCE | 10.0% | 12.0% | 19.9% | 34.3% | 34.2% |
| Net debt / equity | net cash | net cash | net cash | net cash | net cash |
| Net debt / EBIT | net cash | net cash | net cash | net cash | net cash |
| Valuation at $404.40 | |||||
| P/E | 122.2x | 97.0x | 50.3x | 23.2x | 18.3x |
| EV / EBIT | 108.6x | 85.8x | 44.0x | 19.6x | 14.8x |
| P/B | 11.5x | 10.5x | 9.2x | 6.8x | 5.1x |
| FCF yield | 0.4% | 1.0% | 0.7% | 2.9% | 4.8% |
| Dividend yield | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% |
| P/E at target | 138.1x | 109.7x | 56.8x | 26.2x | 20.7x |
Source: Company filings (A); Penmark Research model (E). Non-GAAP EBIT stands in for EBITDA (the model’s EBITDA-equivalent line). ROE, ROA and ROCE on average GAAP capital. Multiples at the 5 Nov 2026 close except the last row. AMD pays no dividend and holds net cash, so dividend and net-debt rows read nil.
| US$m, $/share | Q4’26E | FY26E | FY27E | FY28E |
|---|---|---|---|---|
| Revenue | ||||
| Penmark | 16,964 | 52,130 | 95,336 | 121,319 |
| Consensus | 16,310 | 51,476 | 90,830 | 126,790 |
| Prior year | 10,270 | 34,639 | 52,130 | 95,336 |
| Diff. vs consensus | +4.0% | +1.3% | +5.0% | −4.3% |
| Growth y/y | +65.2% | +50.5% | +82.9% | +27.3% |
| Non-GAAP EPS ($) | ||||
| Penmark | 2.91 | 8.04 | 17.43 | 22.12 |
| Consensus | 2.68 | 7.80 | 16.27 | 23.00 |
| Prior year | 1.53 | 4.17 | 8.04 | 17.43 |
| Diff. vs consensus | +8.6% | +3.1% | +7.2% | −3.8% |
| Growth y/y | +90.1% | +92.9% | +116.7% | +26.9% |
Source: Company filings, guidance of 3 Nov 2026; consensus per InvestingPro, 5 Nov 2026; Penmark Research model. Order: Penmark, consensus, prior-year period, difference, growth.
Source: Company filings (A); Penmark Research model (E); consensus per InvestingPro, 5 Nov 2026.
Next catalyst is NVIDIA on 17 November; AMD trades 66% above peers on P/E but 11% below on PEG using our EPS
| FY2, at 5 Nov price | AMD | Peers | Diff. |
|---|---|---|---|
| Valuation | |||
| P/E, consensus | 24.9x | 15.0x | +66% |
| P/E, our EPS | 23.2x | 15.0x | +55% |
| PEG, consensus | 0.23x | 0.22x | +2% |
| PEG, our EPS | 0.20x | 0.22x | −11% |
| Growth and returns | |||
| EPS growth | 109% | 71% | +38pp |
| Gross margin, TTM GAAP | 50% | 67% | −17pp |
| ROE, TTM GAAP | 8% | 33% | −25pp |
| Debt / market cap | 1.0% | 3.2% | −2.2pp |
| Beta, 5Y, raw | 2.30 | 1.96 | +0.34 |
Source: Company filings, S&P Global and other consensus (7-30 Oct 2026), Nasdaq closes 5 Nov 2026; Penmark Research model. Peers: NVIDIA, Broadcom, Marvell median. Peer beta is the raw 5-year median; the model relevers it to 1.93 at AMD’s leverage. Difference in % for multiples, pp for ratios.
- 17 Nov 2026NVIDIA FQ3 results: AI demand read-across; Bull 1 and Bear 3 context.
- 2 Feb 2027AMD Q4’26: we expect $16.96bn revenue and $2.91 EPS vs consensus $16.31bn and $2.68; FY27 gross margin commentary.
- Feb 2027FY26 10-K: warrant vesting, purchase commitments and Helios customer disclosure.
- May 2027Q1’27: Helios volume and GW shipped; Server CPU doubling on track.
Source: Company calendar, Penmark Research.
- US data-center policy. Permit pauses slow 2027 GW. Likelihood Low (20%); −$74/share.
- Arm CPUs and NVIDIA Vera. Arm and NVIDIA CPUs take sockets. Likelihood Medium (30%); −$41/share.
- Rack mix dilutes margin. Racks pull gross margin below 54.5%. Likelihood Medium (35%); −$18/share.
- Upside, against the cautious view: Helios ramps faster. Lead customers pull volume forward. Likelihood High (60%); +$46/share.
Source: Penmark Research model, catalyst table.
Source: Nasdaq closes via Yahoo Finance; reported non-GAAP quarterly EPS (trailing four quarters); weekly, Jan 2022 to 5 Nov 2026.
Source: Company filings, consensus (7–30 Oct 2026), Nasdaq closes 5 Nov 2026; Penmark Research. NVIDIA, Broadcom, Marvell, Intel, Qualcomm are not covered by Penmark.
Data Center revenue drives the stock, AI is 53% of FY27E growth, and a 24.4x target multiple gives 13% to our target
Source: Company filings (Data Center segment revenue); Nasdaq quarter-end closes via Yahoo Finance; Penmark Research. Eight quarters, 4Q24 to 3Q26.
Source: Penmark Research model, revenue build (US$bn). Server CPU and Data Center AI split is not disclosed; FY26E split 60/40 from management commentary.
Source: Penmark Research model: Data Center AI revenue per GW; OpenAI and Meta warrants (8-Ks, 6 Oct 2025 and 24 Feb 2026). GW, cumulative.
Source: Penmark Research estimates (E); consensus per InvestingPro, 5 Nov 2026. Nine months reported plus Q4 consensus for FY26E.
Source: Penmark Research model; Street targets per stockanalysis.com, 4 Nov 2026. Dots are point values: base, weighted, DCF, Street mean.
Source: Penmark Research model, sensitivity grid; centre is base-case value of $425 at 24.4x and $17.43.
We differ from the market on four debates; the three we can price net to +$11 a share, against a $20 gap to base value
Debates 1 to 3 are priced in the model. Debate 1 is covered in section 10, debate 2 in section 11 and debate 3 in section 12; debate 4 sits in the bear case.
| Debate | Market view | Penmark view | Evidence | What would change our view | Resolves |
|---|---|---|---|---|---|
| 1. Data Center AI growth in FY27E | +137% (value $394). Consensus revenue $90.8bn sits below the 80%+ guide. | +170% to $36.6bn (value $425, +$31). Helios ramps at five lead customers; Q4 guide has Data Center up 50% q/q. | Q3’26 Data Center $8.5bn, +97% y/y; AI passes server in Q4; 2.8GW shipped in FY27E. | Q4’26 Data Center below +45% q/q, or FY27 AI guide at or below 2x. | Feb 2027 |
| 2. FY27E gross margin | 55.6% (value $438). Q3’26 was 56%. | 54.5% (value $425, −$13). Rack-scale Helios carries more third-party content; no FY27 margin guide. | Q4’26 guide is 55% against 56% in Q3; non-GAAP margin was 52% in FY25. | Q4’26 margin of 56% or above with racks already shipping. | Feb 2027 |
| 3. Warrant dilution | No dilution before 2028 (value $432); models use about 1.68bn shares. | 2GW vests in FY27E, 53m more shares and −$7 a share. OpenAI and Meta can take up to 320m shares at $0.01. | Up to 160m shares each, vesting with GW shipped; shipments cover vested GW. | FY26 10-K shows no vesting before the first full GW ships. | Feb 2027 |
| 4. US data-center policy | 5 Nov selloff (−17%) prices a real chance of curbs after the midterms. | We carry it as the largest bear case: −$74 at 20% probability, already inside the 25% bear weight. | Texas grid and permit pauses (Aug, Sep); federal bill S.4214 still in committee. | Passage of a federal bill, or a statewide order in a top-three AI market. | 2027 sessions |
Source: Company reports, S&P Global and InvestingPro consensus (28 Oct-5 Nov 2026), Penmark Research model.
Source: Penmark Research model, Market Check: value per share with one driver at the market view and at ours, others at base.
How the debates map to the thesis
Growth is worth +$31; margin and dilution take back $20. The net of $11 leaves base value $425, 5% above the price; the 13% return to target adds the cost-of-equity roll.
Why the market may be wrong
The price already assumes 148% AI growth. Guidance of 80%+ growth implies at least $93.9bn of FY27E revenue, 3% above consensus; the 17% fall on 5 November was policy fear.
Source: Company calendar and Yahoo Finance earnings dates (2 Feb 2027); the 10-K and later dates are Penmark estimates of the usual reporting cadence.
AMD sells chips into four end markets, and Data Center is moving from 48% of revenue in FY25A to 65% in FY26E
Source: Penmark Research model, FY26E (nine months reported plus Q4 guide); non-GAAP. US$bn. Server CPU and Data Center AI split is not disclosed by AMD; the model splits it 60/40 from management commentary.
Source: Company filings, Q3’26 release (26 Sep 2026 quarter); Penmark Research model.
What it sells, and why customers buy
AMD designs the chips and TSMC builds them. Data Center sells EPYC server CPUs and Instinct accelerators (MI455X in Helios racks) to cloud and AI customers; Client sells Ryzen PC processors; Gaming sells console chips; Embedded sells FPGAs and adaptive chips. Customers buy for performance per watt and a second source to NVIDIA and Intel.
How it is priced
Revenue is units times price times mix. Accelerators are priced per rack or per GW; our build gives about $13bn of Data Center AI revenue per GW shipped.
Cash versus profit
Profit runs ahead of cash in the ramp. FY26E free cash flow is $4.7bn on $15.1bn of operating profit because working capital absorbs $8.2bn. Inventory reached $11.5bn at Q3’26 and unconditional purchase commitments are $48.8bn.
Source: Company filings (FY25 10-K; Q3’26 10-Q commitments at 26 Sep 2026); Penmark Research. AMD is fabless: it does not own wafer fabrication.
Revenue is a tree with two branches that matter: server CPU and Data Center AI explain $43bn of FY27E growth
Source: Company filings (FY25A); Penmark Research model (FY26E). Highlighted boxes are the forecast drivers in sections 10–12 and 14. Server CPU and Data Center AI are model estimates.
Source: Penmark Research model, revenue build, US$bn. Components may not sum due to rounding.
Source: Company filings, Q3’26 release; growth versus Q3’25.
Source: Penmark Research model, FY27E revenue, US$bn. Client $10.3bn (11%), Gaming $2.6bn (3%).
Cost to deliver
Fabless means variable cost of revenue and fixed R&D. Cost of revenue is 44% of FY26E revenue and scales with units; non-GAAP opex of $13.8bn is mostly engineers (about 31,000 staff) and does not. R&D was $8.1bn in FY25A.
Concentration
No customer was 10% of revenue in FY25 or FY24. TSMC supplies wafers, and memory and packaging cost is the supply risk the Street tracks.
The market is growing about 55% a year, but a $51bn FY28E AMD AI line rests on customers who are still raising capex
Source: Global Market Insights (GPU market, excludes custom ASICs); company capex guidance (Amazon, Alphabet, Meta); TrendForce; AMD 8-Ks (6 Oct 2025, 24 Feb 2026); press reports on Ohio. Vendor market sizes differ widely by scope.
Source: Company guidance; Alphabet $195–205bn, Meta $135–145bn, Oracle $90–95bn gross (FY27) shown at midpoints; Amazon about $220bn cash capex. Microsoft excluded: year basis unclear.
Source: Third-party estimates (2025: NVIDIA 80–81%, AMD 6–8%; 2026: NVIDIA 75%, AMD 7–10%), shown at midpoints with other as remainder. Not AMD disclosures; estimates vary by source.
Source: ERCOT and Texas agency notices; congress.gov (S.4214); press reports (3–5 Nov 2026). A broader Texas moratorium was proposed, not issued. Ohio and Maine measures are local, not statewide.
| Vendor | Platform and timing | Why it matters to AMD |
|---|---|---|
| NVIDIA | Vera Rubin NVL72: 72 Rubin GPUs and 36 Vera CPUs per rack | Rack-scale rival to Helios; Vera also contests server CPU sockets |
| AMD | MI455X in Helios racks shipping in volume since Q3’26; MI500 planned for 2027 | Five lead customers: Anthropic, Meta, Microsoft, OpenAI, Oracle |
| Broadcom | Custom accelerators with OpenAI: 10GW target, H2 2026 to end-2029 | Takes AI budget that could go to merchant GPUs |
| Google and AWS | Ironwood TPU now; Trainium4 expected 2027 | In-house silicon at the largest buyers |
| Qualcomm | AI250 sampling expected mid-2027; Dragonfly C1000 in 2028 | Later entrant; limited effect before 2028 |
Source: Company announcements and press reports, Q3’26 call (3 Nov 2026); timing for 2027 and 2028 products is company or press expectation.
AMD has taken about 10 points of x86 server revenue share between Q4’24 and Q1’26, but its AI position still sits behind a 75% margin leader
Source: Third-party x86 server share estimates via press reports; AMD does not disclose share. Only quarters with published estimates are shown (Q4’24, Q4’25, Q1’26, Q2’26); revenue share for Q2 2026 is not available.
Source: Third-party PC CPU unit share estimates via press reports; only quarters with a published figure are shown. Total client unit share was 30.3% in Q2 2026.
| Segment | Main competitors | AMD position | Threat to the thesis |
|---|---|---|---|
| Server CPU | Intel Xeon, Arm CPUs (Graviton, Axion), NVIDIA Vera | Share leader on revenue gains; 540+ public-cloud EPYC instance types | Arm CPUs and Vera: Bear 3, −$41 |
| AI accelerators | NVIDIA (about 75%), Broadcom and Google custom ASICs, Trainium | Second source with five lead customers on MI455X | NVIDIA’s annual cadence and 75% margin |
| Client | Intel, Qualcomm Arm PCs, Apple | 30% unit share; Intel supply limits are a tailwind | Memory costs cut PC units; FY27E client −8% |
| Gaming | Console cycle, NVIDIA discrete graphics | Console chips; 8% discrete add-in share in Q2 2026 | Late-cycle console demand: Q3’26 −40% y/y |
| Embedded | Altera, Microchip, Lattice | Versal; more than $18bn design wins expected in 2026 | Design wins are not booked revenue |
Source: Company filings; third-party share data; Penmark Research. Threat values per share from the model scenarios.
Source: Company results, non-GAAP gross margin. NVIDIA Q3 FY27 guide is 74.0% plus or minus 50bp; AMD Q4’26 guide is about 55%.
Why AMD wins
Customers want a second source and more performance per watt. Turin is in AWS, Google Cloud, Azure and Tencent; Helios gives AI labs an alternative to NVIDIA racks.
What does not last
Share gains in PCs depend on Intel’s supply. We carry Client at −8% in FY27E for that reason.
Management beat its revenue guide in Q1 and Q3 of FY26, and buybacks have covered stock pay; warrants are the new dilution
| Executive | Role | Record | 2026 salary / target equity |
|---|---|---|---|
| Lisa Su | Chair and CEO | CEO since Oct 2014; Chair since Feb 2022 | $1.375m / $36m |
| Jean Hu | EVP, CFO and Treasurer | Joined Jan 2023 | $0.85m / $9m |
| Mark Papermaster | EVP and CTO | Joined Oct 2011 | $0.90m / $10m |
| Forrest Norrod | EVP, Data Center Solutions | Runs the largest segment | $0.80m / $8m |
Source: AMD proxy statement 2026 and Form 8-K; pay is target, not realised.
Source: Company 10-K filings; Penmark Research model. Buybacks as reported in the cash flow statement.
Source: Company releases (Q4’25 on 3 Feb 2026, Q2’26 on 4 Aug 2026); reported results. Bar = reported revenue; grey band = guidance range.
Source: SEC Form 4 filings, Sep–Oct 2026. No open-market purchases, Sep–Oct 2026. Lisa Su also received a 35,000-share grant on 10 Sep 2026.
Governance
Chair and CEO roles are combined. Nora Denzel is lead independent director; seven of eight board nominees are independent.
Pay and dilution
Pay follows per-share outcomes. Long-term awards use relative TSR to August 2029 and an FY28 EPS modifier. The 2026 plan asks for 65m more shares, and customer warrants add up to 320m.
Source: Q3’26 release and 10-Q (26 Sep 2026); FY25 10-K; Penmark Research model.
AMD trades on guidance and the chip sector, not on the quarter: EPS beats moved the price the wrong way in 3 of 8 prints
Four drivers pass at least two of three tests. The stock is a high-beta AI chip proxy, and the reaction to the next guide matters more than the reported EPS.
| Driver | What moves the stock | Tests passed | Value in our model | Debate | Next number |
|---|---|---|---|---|---|
| 1. AI sector demand | SOX explains 26% of variance; AMD fell 17.3% on 5 Nov against 13.9% implied by SOX | Event days; factor test | Policy bear case: −$74 at 20% odds | 4 | NVIDIA FQ3, 17 Nov |
| 2. Data Center AI growth | Price tracks Data Center revenue, correlation 0.91 over eight quarters; the price implies +148% | Price link; return split | +$31 a share vs the market’s +137% | 1 | Q4’26, 2 Feb 2027 |
| 3. Forward guide, not the EPS beat | Reaction to EPS surprise: correlation −0.22; moves of 17% to 19% in Feb and May 2026 | Event days; return split | Multiple fell 12% as EPS rose 78% | 1, 2 | Q4’26, 2 Feb 2027 |
| 4. Gross margin and dilution | Q4 guide of 55% against 56%; warrants vest from 2027 | Event days; return split | −$13 and −$7 a share | 2, 3 | FY26 10-K, Feb 2027 |
Source: Nasdaq closes and EPS surprises via Yahoo Finance; company filings; Penmark Research model. Tests: event days, return split, factor regression.
Source: Nasdaq closes and EPS surprises via Yahoo Finance, results days 4 Feb 2025 to 3 Nov 2026. Average absolute move 7.6%.
Source: Nasdaq closes via Yahoo Finance to 5 Nov 2026; trailing four-quarter non-GAAP EPS from company releases ($6.65 now; $3.73 a year ago; $2.57 three years ago). No dividend is paid.
Source: Penmark Research regression, 157 weekly returns, Nov 2023 to Nov 2026; S&P 500 alone, then SOX, NVIDIA-Broadcom-Marvell basket and 10-year yield added in turn. Total R-squared 0.54; the 10-year yield adds 0.0%.
What the tests say
The market sets the tone, the sector sets the size. The S&P 500 and SOX explain 53% of weekly moves; the other 46% is company news, mostly guidance.
AMD fell 17.3% on 5 November, 3.3pp more than its chip-sector beta implied (z-score −1.4, not extreme).
The drivers we keep are worth +$31 to −$74 a share; rates, peers and EPS beats do not move the price
Source: Penmark Research model, Market Check and Thesis tabs; first three are market view against ours, policy is the bear case at 20% probability. US$ per share.
| Candidate | Result | Verdict |
|---|---|---|
| 10-year Treasury yield | R-squared 0.02% after market, sector and peers | Dropped |
| AI peer basket (NVIDIA, Broadcom, Marvell) | 0.8% beyond SOX, which already holds them | Dropped |
| EPS beat size | Correlation −0.22 with the next-day move | Dropped |
| Consensus revisions | FY27E EPS up 17% in three months; the stock fell 36% in a month | Dropped |
Source: Penmark Research regression and event study; Yahoo Finance.
What the market does: SOX explains 26% of variance beyond the market, and AMD fell 17.3% on 5 Nov on data-center policy fears. Our model: a 2027 policy limit costs $74 a share at 20% probability. Next number: NVIDIA FQ3 on 17 Nov 2026; our estimate is not set, as NVIDIA is not covered.
What the market does: price implies +148% against our +170%. Our model: value $425 against $394 at the market’s +137%; the gap to the price is $20. Next number: Q4’26 on 2 Feb 2027; we model $16.96bn revenue and $2.91 EPS against consensus $16.31bn and $2.68.
What the market does: a 15.5% EPS surprise in Feb 2026 met a 17.3% fall; a 5.8% surprise in May met a gain of 18.6%. Our model: the FY27 guide is already above 80% growth and we are at 83%. Next number: FY27 Data Center AI guide of “significantly more than double”; average absolute move on results is 7.6%.
What the market does: consensus implies 55.6% gross margin and about 1.68bn shares. Our model: 54.5% and 1.70bn average shares. Next number: Q4’26 gross margin guided about 55%; FY26 10-K on warrant vesting.
Data Center AI more than doubles in FY27E because lead customers have committed and Helios is shipping
What the market believes
Consensus has FY27E revenue of $90.8bn, below the 80% growth AMD guided on 3 November. The price implies +148% Data Center AI growth, or $92.3bn of revenue, against our +170% and $95.3bn.
The evidence
- Reported data. Data Center revenue was $8.5bn in Q3’26, +97% y/y; the Q4 guide of +50% q/q implies $12.8bn.
- Filings. OpenAI and Meta each hold a 1GW firm commitment for MI450 and warrants on up to 160m shares; our build ships 2.8GW in FY27E.
- Call statements. AMD named Anthropic, Meta, Microsoft, OpenAI and Oracle as Helios lead customers and guided Data Center AI to “significantly more than double”.
- Industry data. Amazon, Alphabet and Meta guide to $560bn of 2026 capex, and 2027 is expected higher.
Source: Company filings and Q3’26 release; segment revenue, US$m. Q4’26 guide is +50% q/q, or about $12.8bn.
| US$m | FY26E | FY27E | FY28E |
|---|---|---|---|
| Revenue, Penmark | $52,130 | $95,336 | $121,319 |
| Revenue, consensus | $51,476 | $90,830 | $126,790 |
| Difference | +1.3% | +5.0% | −4.3% |
What it is worth
Each 10pp of Data Center AI growth is worth about $9 a share. Our +170% gives a value of $425; the market’s +137% gives $394. This is the largest single gap to consensus and drives 53% of FY27E revenue growth.
Signposts
- 2 Feb 2027. Q4’26: Data Center near $12.8bn confirms; below +45% q/q disproves.
- Feb 2027. FY27 AI guide at more than 2x supports; at or below 2x breaks the claim.
- Mid-2027. Q1 and Q2’27 GW shipped against 2.8GW for FY27E.
Source: Penmark Research model, Scenarios and Market Check tabs. Price-implied growth holds the target P/E and other drivers at base.
Operating leverage lifts the operating margin to 36% even though rack-scale Helios dilutes gross margin
What the market believes
Consensus implies a 55.6% FY27E gross margin and about 1.68bn shares. The Q4 guide of 55% is a point below Q3’s 56%, and the CFO did not guide FY27 margin.
The evidence
- Reported data. Non-GAAP gross margin was 55.9% in Q3’26 and 56.2% in Q2’26 against 52% in FY25; operating margin was 29.3% in Q3’26.
- Call statements. The CFO expects “very significant operating margin expansion in 2027”; Q4 opex is guided at about $3.75bn, up 5% q/q on revenue up 27%.
- Long-term targets. The November 2025 analyst day set FY28 gross margin at 55% to 58% and operating margin above 35%; we model 55.0% and 37.2%.
- Peer data. NVIDIA earns a 75% gross margin on rack-scale systems, so the ceiling is higher than AMD shows today.
Source: Company releases, non-GAAP gross profit over revenue; Q2’25 includes an $800m MI308 inventory charge; Q4’26 guide is about 55%.
| FY27E | Penmark | Consensus | Diff. |
|---|---|---|---|
| Revenue ($m) | 95,336 | 90,830 | +5.0% |
| Gross margin | 54.5% | 55.6% | −1.1pp |
| Average diluted shares (m) | 1,700 | 1,680 | +1.2% |
| Non-GAAP EPS | $17.43 | $16.27 | +7.2% |
What it is worth
Margin is a $13 drag, leverage is the offset. At the consensus margin our value would be $438, $13 above base. Revenue above consensus adds $1.18 to FY27E EPS, margin takes off $0.54 and shares $0.25; opex and other lines add the remaining $0.77 of the $1.16 gap.
Signposts
- 2 Feb 2027. Q4’26 margin of 56% or above with racks shipping would confirm the upside.
- Q1’27. First full quarter of Helios mix: below 53% for two quarters breaks the claim.
- FY26 10-K. Opex growth against the 30% we model.
Source: Penmark Research model, Scenarios tab; consensus-implied margin per Market Check tab. Per-share value from the sensitivity grid.
Customer warrants dilute EPS by a visible 3% a year, and every vested share is paid for by shipments
What the market believes
Most models use about 1.68bn shares and no warrant dilution. We use 1,700m average shares in FY27E and 1,761m in FY28E because vesting follows GW shipped.
The evidence
- Filings. OpenAI and Meta can each take up to 160m shares at $0.01; vesting is tied to GW shipped, up to 6GW each.
- Reported data. Q3’26 GAAP net income took a $1.1bn warrant charge, excluded from non-GAAP EPS.
- Model test. Implied shipments of 2.8GW cover the 2GW vested in FY27E, and cover holds every year.
- Proxy. The 2026 plan seeks 65m extra employee shares on top of the warrants.
Source: Penmark Research model, Statements tab; year-end diluted shares. OpenAI and Meta warrants at $0.01, up to 160m shares each (8-Ks, 6 Oct 2025 and 24 Feb 2026); model vests 2GW a year to a modelled 160m total by FY29E.
| Warrant schedule | FY27E | FY28E | FY29E |
|---|---|---|---|
| GW vested, cumulative | 2.0 | 4.0 | 6.0 |
| GW shipped, cumulative | 2.8 | 6.8 | 11.7 |
| Warrant shares vested (m) | 53 | 107 | 160 |
| Cover: shipped over vested | 1.4x | 1.7x | 1.9x |
What it is worth
Dilution costs $7 a share, a 1.6% haircut to value. It is the smallest of the three debates but the surest: the warrants are contractual and vest as the revenue we forecast arrives.
Signposts
- Feb 2027. FY26 10-K: first vesting tranche conditions and share count.
- Q1’27. Q1 diluted share count; our 1,700m FY27E average implies about 1,690m.
- 2027. Any change in the 6GW tranche terms.
Source: Company 10-K filings (A); Penmark Research model (E). FY22 includes the Xilinx acquisition shares.
Revenue has grown 20% a year since FY21A, but cash conversion swings with working capital and non-GAAP excludes $5bn of real costs
Source: Company 10-K filings. FY22 includes the Xilinx acquisition (completed February 2022).
Source: Company filings (A); Penmark Research model (E). GAAP EPS includes amortisation of acquired intangibles and warrant fair-value changes, warrant fair-value changes are not forecast.
Source: Company filings (A); Penmark Research model (E). Free cash flow over non-GAAP net income, %.
| Item | Treatment and size |
|---|---|
| Amortisation of acquired intangibles | $2.2bn in FY26E excluded from non-GAAP; non-cash, but it is the cost of buying growth. |
| Stock-based compensation | $1.6bn in FY25A (4.7% of revenue); excluded from non-GAAP EPS, treated as a cash cost in our DCF. |
| Warrant fair value | $1.1bn charge in Q3’26 excluded from non-GAAP; the dilution is in our share count. |
| Inventory charge | $800m MI308 charge in Q2’25 cut that quarter’s non-GAAP margin to 43%. |
| Tax | FY25 GAAP tax was a $103m benefit on reserve releases; non-GAAP rate is 13% to FY27E and 15% from FY28E. |
Source: Company filings and releases; Penmark Research model.
Our FY27E revenue is 5% above consensus and 1.5% above the guided floor, and EPS is 7% above on two drivers
Source: Company filings; Penmark Research model, FY27E. Highlighted boxes are the two forecast drivers. Server CPU and Data Center AI split is a model estimate.
| Line | Penmark | Consensus | Guidance | Gap | Reason |
|---|---|---|---|---|---|
| Revenue FY26E ($m) | 52,130 | 51,476 | 51,656 to 52,676 | +1.3% | Q4 built from segment guidance; Data Center +50% q/q |
| Revenue FY27E ($m) | 95,336 | 90,830 | above 93,899 | +5.0% | Guide is above 80% growth; consensus has not caught up |
| Revenue FY28E ($m) | 121,319 | 126,790 | CAGR above 35% | −4.3% | Server and AI growth slow to 25% and 40%; consensus is higher |
| Gross margin FY27E | 54.5% | 55.6% | none | −1.1pp | Rack-scale mix; no FY27 margin guide |
| Gross margin FY28E | 55.0% | – | 55% to 58% | – | Low end of the analyst-day range |
| EPS FY26E | $8.04 | $7.80 | none | +3.1% | Revenue beat on Q4 guide midpoint |
| EPS FY27E | $17.43 | $16.27 | none | +7.2% | Revenue +$1.18, margin −$0.54, shares −$0.25, opex +$0.77 |
| EPS FY28E | $22.12 | $23.00 | above $20 | −3.8% | Warrant shares reach 107m; consensus revenue higher |
Source: Company guidance (3 Nov 2026) and analyst-day targets (11 Nov 2025); consensus per InvestingPro and S&P Global, 5 Nov 2026; Penmark Research model. Consensus gross margin is implied, not published.
Source: Company guidance (3 Nov 2026); consensus per InvestingPro, 5 Nov 2026; Penmark Research model. EPS is not guided. Bar = Penmark Q4’26 estimate; grey band = guidance range; amber tick = consensus.
Value moves most with the target multiple and least with margin; free cash flow lags profit until FY27E as inventory builds
| US$m | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|
| Operating cash flow | 7,709 | 7,195 | 23,859 | 37,036 |
| Capital expenditure | (974) | (2,545) | (4,290) | (4,853) |
| Free cash flow | 6,735 | 4,650 | 19,568 | 32,183 |
| Change in working capital | – | (8,222) | (7,597) | (4,687) |
| FCF margin | 19.4% | 8.9% | 20.5% | 26.5% |
| Capex % of revenue | 2.8% | 4.9% | 4.5% | 4.0% |
| Inventories ($m) | 7,920 | 14,798 | 20,931 | 23,623 |
| Receivables ($m) | 6,315 | 12,524 | 18,401 | 22,446 |
Source: Company filings (A); Penmark Research model (E).
Costs and margins
Cost of revenue varies with units; opex is mostly fixed. FY27E non-GAAP opex is $18.0bn, up 30% on revenue up 83%, so operating margin rises to 35.6%. Gross margin falls to 54.5% on rack mix.
Cash and balance sheet
Working capital absorbs $8.2bn in FY26E. Inventory reaches $14.8bn and receivables $12.5bn. Cash builds to $35.3bn including investments by FY27E and debt stays at $7.1bn.
Source: Penmark Research model, Scenarios tab sensitivity grids; value per share at the price date, one input moved with others at base.
Source: Penmark Research model, Scenarios tab; value per share at the price date. Centre is the base case, $425.
Source: Company filings (A); Penmark Research model (E). Non-GAAP gross and operating margin.
Source: Company filings (A); Penmark Research model (E). The model assumes no buybacks or acquisition cash outflow, so the cash build is a ceiling.
Forward P/E on a peer PEG gives a $425 base value and a $457 target; the DCF at a 13% discount rate is the cross-check
Method
AMD is profitable and earnings are ramping, so we value it on FY27E P/E. The multiple is 24.4x: the median PEG of NVIDIA, Broadcom and Marvell (0.22x) times AMD’s consensus EPS growth of 109%. EV/EBITDA adds nothing for a net-cash company, and sum of the parts does not fit one fabless model.
Rating
Total return of 13.1% is Buy, with low conviction. One of eight robustness tests flips the rating: the lowest AI peer PEG (0.16x) gives $329 and −19%.
Methods compared
The P/E methods span $276 to $510 and the DCF sits far below at $188 to $277. The football field is Exhibit 21 on page 5. The gap is the discount rate: the P/E multiple embeds the market’s pricing of peers, while the DCF uses a 13.0% cost of capital.
Street
The Street range is $430 to $1,250, with a mean of $649. Our $457 target is $27 above the lowest Street target.
Source: Penmark Research model, Valuation tab. No dividends. Total return to target is 13.1% against a 13.0% cost of equity.
| Multiple | At $404.40 | At target $457.28 |
|---|---|---|
| P/E, FY27E | 23.2x | 26.2x |
| P/E, FY28E | 18.3x | 20.7x |
| Target forward P/E | 24.4x | 24.4x before roll |
| Peer median PEG | 0.22x | 0.22x |
| AMD EPS growth used | 109% | 109% |
Source: Penmark Research model, Valuation tab. Multiples at the actual price are used elsewhere in this report.
| Company | Price, 5 Nov | P/E, FY2 | EPS growth | PEG | In target set |
|---|---|---|---|---|---|
| NVIDIA | $181.60 | 11.4x | 70.9% | 0.16x | Yes |
| Broadcom | $291.53 | 15.0x | 66.8% | 0.22x | Yes |
| Marvell | $133.55 | 18.2x | 73.7% | 0.25x | Yes |
| Intel | $52.25 | 35.8x | 39.0% | 0.92x | No |
| Qualcomm | $156.75 | 15.4x | −2.7% | n/m | No |
| AI compute median | 15.0x | 70.9% | 0.22x | ||
| AMD, consensus | $404.40 | 24.9x | 108.6% | 0.23x | |
| AMD, our EPS | $404.40 | 23.2x | 116.7% | 0.20x |
Source: Company filings, consensus (7-30 Oct 2026), Nasdaq closes 5 Nov 2026; Penmark Research. FY2 is the second forecast fiscal year; fiscal year ends differ. Peers are not covered by Penmark.
The DCF gives $223 because it uses a 13.0% discount rate; at the 8.7% the price implies it gives $404
| DCF, US$m | Value |
|---|---|
| PV of FY26E stub to FY36E cash flows | 245,767 |
| PV of terminal value | 183,475 |
| Enterprise value | 429,241 |
| Terminal value share of EV | 42.7% |
| Plus cash and investments | 15,776 |
| Less debt | (7,079) |
| Equity value | 437,938 |
| Diluted shares incl. warrants (m) | 1,990 |
| Value per share at price date | $223.05 |
Source: Penmark Research model, Valuation tab. Equity value of $438bn over 1,990m shares is $220.07 at 26 Sep 2026; rolled forward 0.11 years at the cost of equity to the 5 Nov price date it is $223.05. Terminal growth 3%; ten full years to FY36E.
Source: Penmark Research model, reverse DCF on the Valuation tab; model cash flows and 3% terminal growth held. WACC tested from 6.0% to 15.0%.
Source: Penmark Research model, Scenarios tab; value per share at the price date. Centre $223.
Source: Penmark Research model, Scenarios tab, values at the price date before the one-year roll. Probabilities are analyst judgements: bear 25% reflects policy risk and an unguided FY27 margin.
Why the two methods differ by 48%
The gap is the discount rate, not the cash flows. The DCF equals the P/E value at a WACC of 8.4%, or a beta of 0.84, against the 1.93 we use from peers (AMD’s own 5-year beta is 2.3).
| Case | Probability | FY27E revenue | FY27E EPS | Value | Story and key assumptions |
|---|---|---|---|---|---|
| Bear | 25% | $76.7bn | $11.31 | $276 | Policy limits US build; rack margin 53.0%; Arm CPUs take sockets |
| Base | 55% | $95.3bn | $17.43 | $425 | FY27 framework delivered; warrants vest with GW |
| Bull | 20% | $105.1bn | $20.91 | $510 | Helios ramps faster at five lead customers; Venice beats guide |
Source: Penmark Research model, Scenarios tab. Probability-weighted value is $404.51; probabilities are analyst judgements and do not carry statistical confidence.
The February 2027 results and 10-K decide three of four debates, and we are 4% above consensus on revenue into them
Source: Company calendar; Yahoo Finance earnings dates (2 Feb 2027); the 10-K and later quarterly dates are Penmark estimates of the usual cadence.
| Date | Event | Penmark against consensus | Value at stake | Rating effect |
|---|---|---|---|---|
| 17 Nov 2026 | NVIDIA FQ3 results | Read-across on AI demand and rack supply; not covered by Penmark | Sector tone; ±5% on the day is typical | Neutral |
| 2 Feb 2027 | Q4’26 results: we expect $16.96bn revenue and $2.91 EPS; consensus $16.31bn and $2.68 | Revenue 4.0% and EPS 8.6% above consensus; guide midpoint $17.0bn | Bull 1 (+$46) or Bear 2 (−$18) | Supports |
| Feb 2027 | FY26 10-K and FY27 guide: Data Center AI, margin, first warrant vesting | We model AI +170%, margin 54.5%, 2GW vested | Debates 1 to 3: +$31, −$13, −$7 | Could change rating |
| 2027 | State and federal data-center policy after the 3 Nov midterms | We carry a 20% chance of limits on 2027 GW | Bear 1: −$74 a share | Could change rating |
Source: Company releases and calendar; consensus per InvestingPro, 5 Nov 2026; Penmark Research model.
Source: Penmark Research model, Thesis tab; US$ per share if the catalyst lands, before probability weights shown.
Events that could change the rating
A Q4 guide below 45% Data Center growth or a policy limit would end the Buy. The rating already flips at the lowest peer PEG, so conviction is low. A confirmed 56% margin and an AI guide above 2x would lift the case towards the bull value of $510.
Source: Company guidance (3 Nov 2026), consensus per InvestingPro (5 Nov 2026), Penmark Research model. FY27E guide is a floor (growth above 80%). Bar = Penmark estimate; grey band = guidance; amber tick = consensus.
Next number to watch
| Driver | When | Our estimate |
|---|---|---|
| Data Center AI growth | Q4’26, 2 Feb 2027 | Data Center $12.8bn, +50% q/q |
| Gross margin | Q4’26, 2 Feb 2027 | 55.0% (guide about 55%) |
| Warrant dilution | FY26 10-K, Feb 2027 | 0 vested at FY26; 2GW by FY27E |
| Policy | 2027 sessions | No limit in base case |
Five downside risks cost $7 to $74 a share; the thesis ends if Data Center AI growth falls below the +148% in the price
| Risk | What happens | Likelihood | Value | Signpost to watch |
|---|---|---|---|---|
| Bear 1: US data-center policy | Permit pauses and a federal moratorium slow 2027 GW deployed | Low (20%) | −$74 | Federal bill S.4214 moves; statewide order in a top-three AI market |
| Bear 3: Arm CPUs and NVIDIA Vera | Server CPU growth is +70% not +100% | Medium (30%) | −$41 | Venice share data; hyperscaler Arm launches |
| Bear 2: Rack mix dilutes margin | Gross margin 53.0% not 54.5% | Medium (35%) | −$18 | Q4’26 margin below 55%; FY27 margin commentary |
| Bear 4: PC decline deepens | Client −18% not −8% on memory costs | Medium (40%) | −$8 | Intel supply; memory prices; 1H27 sell-through |
| Bear 5: Warrants vest as modelled (market assumes none) | 2GW vested in FY27E: 53m shares | High (70%) | −$7 | FY26 10-K vesting conditions |
Source: Penmark Research model, Thesis tab; value per share if the risk occurs, before probability weights.
Source: Penmark Research model, Thesis tab catalyst values and probabilities.
What would end the thesis
Data Center AI growth below +148% for FY27E. That is the growth the price already implies; below it our value falls under the price. The rating also falls to Sell if the target PEG drops to the lowest peer level of 0.16x ($329, −19%).
Regulatory, legal, key-person
Policy and export rules matter most. Lisa Su is Chair and CEO; no succession plan is disclosed in AMD’s filings. Insider sales of $40.6m were under 10b5-1 plans.
Source: Penmark Research model, Thesis tab: value per share times analyst-judged probability. Risks overlap, so the sum ($41) is not additive to the target.
Index funds and large managers own 75% of AMD, shorts are 2.3% of float, and the Street is 83% Buy
Source: Yahoo Finance holders and short-interest data (13F as of 30 Jun 2026); Nasdaq price history to 5 Nov 2026.
| Holder | % held | Shares (m) | Change |
|---|---|---|---|
| BlackRock | 9.2% | 150.3 | +3.3% |
| Vanguard Capital Mgmt | 6.5% | 106.5 | +0.5% |
| State Street | 4.5% | 74.4 | −0.6% |
| Invesco | 3.2% | 53.2 | +5.4% |
| JPMorgan Chase | 2.5% | 40.3 | +66.2% |
| Geode Capital | 2.4% | 39.6 | −0.6% |
| Vanguard Portfolio Mgmt | 2.3% | 37.9 | +0.9% |
| T. Rowe Price | 2.2% | 35.9 | +24.6% |
| FMR (Fidelity) | 1.6% | 25.6 | +67.2% |
| Morgan Stanley | 1.5% | 24.3 | +2.1% |
Source: Form 13F filings via Yahoo Finance, quarter to 30 Jun 2026. Change is in shares held versus the prior filing.
Source: TipRanks, 4 Nov 2026 (ratings in the last three months); no Sell ratings.
Source: stockanalysis.com, 4 Nov 2026; Penmark Research model.
Source: Nasdaq closes via Yahoo Finance, 5 May to 5 Nov 2026; SOX rebased to the AMD price on 5 May 2026. Peak $658.52 on 6 Oct 2026; 52-week low $188.22 on 3 Mar 2026.
Overhangs
Share supply is rising from three sources. Warrants for up to 320m shares to OpenAI and Meta; about $8.2bn of World Labs consideration paid in stock; and a request for 65m more employee shares. We model 1,761m average shares in FY28E.
Insiders and liquidity
Executives sold $40.6m under 10b5-1 plans and made no open-market purchases. Planned sales are not a signal on their own. Daily value traded of $11.6bn means position size is not a constraint.
Revenue rises 3.5 times from $34.6bn to $121.3bn by FY28E as non-GAAP EPS rises from $4.17 to $22.12
| Income statement | FY24A | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|---|
| Revenue | 25,785 | 34,639 | 52,130 | 95,336 | 121,319 |
| Growth y/y | 13.7% | 34.3% | 50.5% | 82.9% | 27.3% |
| Cost of revenue | (12,026) | (16,474) | (23,144) | (43,378) | (54,594) |
| Gross profit | 13,759 | 18,165 | 28,986 | 51,958 | 66,726 |
| Gross margin | 53.4% | 52.4% | 55.6% | 54.5% | 55.0% |
| Operating expenses | (7,621) | (10,397) | (13,849) | (18,004) | (21,604) |
| Operating income | 6,138 | 7,768 | 15,137 | 33,954 | 45,121 |
| Operating margin | 23.8% | 22.4% | 29.0% | 35.6% | 37.2% |
| Net income | 5,420 | 6,831 | 13,354 | 29,640 | 38,950 |
| Diluted shares (m) | 1,637 | 1,636 | 1,660 | 1,700 | 1,761 |
| EPS ($) | 3.31 | 4.17 | 8.04 | 17.43 | 22.12 |
| EPS growth | – | 26.0% | 92.9% | 116.7% | 26.9% |
| GAAP EPS ($) | 1.00 | 2.65 | 6.04 | 14.93 | 19.69 |
| Cash flow | FY24A | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|---|
| Operating cash flow | 3,041 | 7,709 | 7,195 | 23,859 | 37,036 |
| Capital expenditure | (636) | (974) | (2,545) | (4,290) | (4,853) |
| Free cash flow | 2,405 | 6,735 | 4,650 | 19,568 | 32,183 |
| FCF margin | 9.3% | 19.4% | 8.9% | 20.5% | 26.5% |
| Stock-based compensation | 1,407 | 1,638 | 2,058 | 2,860 | 3,276 |
| Share repurchases | (862) | (1,316) | – | – | – |
| Change in working capital | – | – | (8,222) | (7,597) | (4,687) |
| Balance sheet | FY24A | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|---|
| Cash and investments | 5,132 | 10,552 | 15,748 | 35,317 | 67,500 |
| Receivables | 6,192 | 6,315 | 12,524 | 18,401 | 22,446 |
| Inventories | 5,734 | 7,920 | 14,798 | 20,931 | 23,623 |
| Goodwill and intangibles | 43,769 | 41,831 | 40,042 | 38,002 | 36,248 |
| Total assets | 69,226 | 76,926 | 102,373 | 136,926 | 178,236 |
| Debt | 1,721 | 3,222 | 7,079 | 7,079 | 7,079 |
| Shareholders' equity | 57,568 | 62,999 | 72,858 | 101,095 | 139,045 |
| Net cash | 3,411 | 7,330 | 8,669 | 28,238 | 60,421 |
| Book value per share ($) | 35.17 | 38.51 | 43.88 | 59.46 | 78.95 |
Quarterly revenue rises from $7.7bn to a modelled $17.0bn in Q4’26 as Data Center grows 3.3 times
| US$m | Q4’24A | Q1’25A | Q2’25A | Q3’25A | Q4’25A | Q1’26A | Q2’26A | Q3’26A | Q4’26E |
|---|---|---|---|---|---|---|---|---|---|
| Revenue | 7,658 | 7,438 | 7,685 | 9,246 | 10,270 | 10,253 | 11,536 | 13,377 | 16,964 |
| Data Center | 3,859 | 3,674 | 3,240 | 4,341 | 5,380 | 5,775 | 6,718 | 8,546 | 12,819 |
| Client and Gaming | 2,876 | 2,941 | 3,621 | 4,048 | 3,940 | 3,605 | 3,841 | 3,722 | 2,903 |
| Embedded | 923 | 823 | 824 | 857 | 950 | 873 | 977 | 1,109 | 1,242 |
| Gross profit | 4,140 | 3,992 | 3,326 | 4,992 | 5,855 | 5,685 | 6,488 | 7,483 | 9,330 |
| Operating income | 2,026 | 1,779 | 897 | 2,238 | 2,854 | 2,540 | 3,094 | 3,923 | 5,580 |
| Net income | 1,777 | 1,566 | 781 | 1,965 | 2,519 | 2,265 | 2,760 | 3,471 | 4,858 |
| EPS ($) | 1.09 | 0.96 | 0.48 | 1.20 | 1.53 | 1.37 | 1.66 | 2.09 | 2.91 |
| Gross margin | 54.1% | 53.7% | 43.3% | 54.0% | 57.0% | 55.4% | 56.2% | 55.9% | 55.0% |
| Operating margin | 26.5% | 23.9% | 11.7% | 24.2% | 27.8% | 24.8% | 26.8% | 29.3% | 32.9% |
Source: Company releases (A); Penmark Research model (E). Q4’26E is built from the 3 Nov 2026 guidance; later quarters are not modelled quarterly.
| Segment | FY24A | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|---|
| Server CPU (model est.) | – | – | 20,315 | 40,630 | 50,787 |
| Data Center AI (model est.) | – | – | 13,543 | 36,567 | 51,193 |
| Data Center | 12,579 | 16,635 | 33,858 | 77,196 | 101,980 |
| Client | 7,054 | 10,640 | 11,196 | 10,301 | 10,713 |
| Gaming | 2,595 | 3,910 | 2,875 | 2,587 | 2,587 |
| Embedded | 3,557 | 3,454 | 4,201 | 5,251 | 6,039 |
| Total revenue | 25,785 | 34,639 | 52,130 | 95,336 | 121,319 |
Source: Company filings (A); Penmark Research model (E). Server CPU and Data Center AI are model estimates; FY24A to FY25A Client and Gaming as recast.
| Ratios | FY24A | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|---|
| ROE | 2.9% | 7.2% | 14.8% | 29.2% | 28.9% |
| ROA | 2.4% | 5.9% | 11.2% | 21.2% | 22.0% |
| ROCE | 10.0% | 12.0% | 19.9% | 34.3% | 34.2% |
| P/E at $404.40 | 122.2x | 97.0x | 50.3x | 23.2x | 18.3x |
| EV/operating income | 108.6x | 85.8x | 44.0x | 19.6x | 14.8x |
| P/B | 11.5x | 10.5x | 9.2x | 6.8x | 5.1x |
| FCF yield | 0.4% | 1.0% | 0.7% | 2.9% | 4.8% |
| P/E at target $457.28 | 138.1x | 109.7x | 56.8x | 26.2x | 20.7x |
| Net cash / equity | 5.9% | 11.6% | 11.9% | 27.9% | 43.5% |
Source: Company filings (A); Penmark Research model (E). EV uses net cash of $8.7bn at 26 Sep 2026; no dividend is paid, so there is no dividend yield.
Reference: one primary method, one cross-check, a 5 November 2026 data cut-off and a stated basis of presentation
| Adjusted to reported | Non-GAAP EPS $4.17 in FY25A against GAAP $2.65: the $1.52 gap is amortisation of acquired intangibles, stock-based compensation, restructuring and tax effects. |
| Business model | Products and units, fabless semiconductors: revenue is units times price times mix by segment. Data Center is split into server CPU and Data Center AI by the model, not by AMD. |
| Primary valuation | Forward P/E on FY27E non-GAAP EPS at a peer-PEG multiple of 24.4x. Target is probability-weighted value rolled forward one year at the 13.0% cost of equity, no dividends. |
| Cross-check | DCF: Q4’26 stub and FY27E to FY36E, mid-period discounting from 26 Sep 2026, SBC treated as a cash cost, 3% terminal growth, 1,990m shares including warrants. |
| Rating bands | Buy above +10%, Hold between −10% and +10%, Sell below −10% expected total return. High conviction needs a return above 20% and a rating that survives every robustness test. |
| Fiscal year | Ends the last Saturday of December: FY26 is 28 Dec 2025 to 26 Dec 2026. FY26E is nine months reported plus a Q4 built from guidance. |
| Basis of presentation | Price, consensus and peer data are at the 5 Nov 2026 close; the model date is 6 Nov 2026. Non-GAAP operating income is used where the template shows EBITDA; adjusted EBITDA is $16.1bn in FY26E and $35.1bn in FY27E. Consensus EBITDA is not published. AMD’s own forward P/E history is not shown; Exhibit 15 uses trailing P/E. |
| GW | Gigawatt of AI accelerator capacity; our build uses about $13bn of Data Center AI revenue per GW. |
| Helios, MI455X | AMD rack-scale AI system and its accelerator, in volume from Q3’26. |
| EPYC (Venice, Turin) | AMD server CPU brands and generations. |
| PEG | P/E divided by EPS growth in percent, here FY2 P/E over FY1-to-FY2 growth. |
| Non-GAAP | AMD basis excluding amortisation of acquired intangibles, stock-based compensation, warrant fair value and other items. |
| Cost of equity | Risk-free 4.9% plus beta 1.93 times equity risk premium 4.23%, giving 13.05%. |
| Company data | AMD 10-K and 10-Q filings, Q3’26 release and call (3 Nov 2026), proxy statement 2026; company filings data. |
| Market data | Nasdaq closes via Yahoo Finance to 5 Nov 2026; holders and short interest as of the latest Yahoo Finance data. |
| Consensus | InvestingPro (5 Nov 2026), S&P Global via stockanalysis.com (28 Oct 2026), TipRanks (4 Nov 2026). |
| Industry data | Company capex guidance; third-party share and market-size estimates (Global Market Insights, TrendForce and CPU unit and revenue share trackers). These are estimates, not AMD disclosures. |
Analyst certification and important disclosures
Ratings
Ratings reflect the expected total return to our 12-month price target, including dividends, weighted across bull, base and bear scenarios.
| Buy | Expected total return above +10% |
| Hold | Expected total return between −10% and +10% |
| Sell | Expected total return below −10% |
We review ratings at each set of results and whenever the facts change materially. A share price can move outside its band before the rating changes, and targets may not be reached.
Analyst certification
The Penmark Team certifies that (1) the views in this report accurately reflect its views about the subject company and its securities, and (2) no part of its compensation was, is or will be directly or indirectly related to the specific ratings, price targets or views in this report.
Contact
Penmark Research
Questions about our research:
research@penmarkresearch.com
All our research and coverage:
penmarkresearch.com
Basis of preparation
This report uses information publicly available up to its publication date. Share prices are closing prices on the trading day before the publication date, unless stated otherwise. Forecasts, valuations and price targets are Penmark Research’s own, and forecast periods are marked E. Sources are shown beneath each exhibit.
Valuation and price targets
Price targets are based on the valuation methods set out in each report and are weighted across bull, base and bear scenarios. The main risks to each rating and price target are described in the report. Penmark Research updates its coverage after each set of results but is not obliged to update a report at any other time.
Independence and conflicts of interest
Penmark Research is independent. It does not provide investment banking, brokerage, market-making or asset management services, does not trade securities, and does not accept payment from any company for research coverage. Companies do not see or review Penmark Research reports before publication.
No member of the Penmark Team holds a long or short position, directly or through derivatives, in the securities of any company Penmark Research covers.
Regulatory status
Penmark Research publishes general, impersonal investment research on a regular basis. It is not registered as an investment adviser with the US Securities and Exchange Commission or any state securities authority, is not a registered broker-dealer and is not a member of FINRA. Penmark Research does not give advice tailored to any person’s circumstances.
Not investment advice or an offer
This report is for information only. It is not personalised investment advice and is not an offer or solicitation to buy or sell any security or other financial instrument. It does not take into account any reader’s objectives, financial situation or needs. Readers should make their own investment decisions and seek independent professional advice where appropriate.
Forward-looking statements and risk
Forecasts, valuations and price targets are estimates based on assumptions that may prove wrong, and they may not be achieved. Opinions are current as of the publication date and may change without notice. Investing in equities involves risk, including the loss of capital. Share prices can be volatile, and past performance is not a reliable indicator of future results.
Data and sources
Company financials, operating data and filings come from company filings, earnings releases and earnings calls. Market data and any consensus estimates come from third-party sources named beneath the relevant exhibit; consensus figures from aggregators are unofficial. Penmark Research does not use other brokers’ research. Third-party data belong to their providers and may not be extracted from this report.
Accuracy and liability
Information is taken from sources believed to be reliable, but Penmark Research makes no representation or warranty as to its accuracy, completeness or timeliness. To the fullest extent permitted by law, Penmark Research accepts no liability for any loss arising from use of this report. Descriptions of litigation and regulatory matters reflect allegations and public records, not findings, unless stated. If a published report is corrected, the corrected version carries an “Updated” date at penmarkresearch.com.
Distribution and access
Penmark Research reports are available to registered users of penmarkresearch.com. Downloads and programmatic access are governed by the Penmark Research subscription terms. This report is not directed at, or intended for, any person in Singapore or India, and access from those countries is blocked. It is not intended for distribution to, or use by, any person in any jurisdiction where that would be contrary to law or regulation or would subject Penmark Research to registration or licensing requirements there.
Copyright and permitted use
© Penmark Research. All rights reserved. Penmark Research is a trade name of Brightriver Labs. Readers may share a link to this report but may not copy, reproduce, redistribute, sell or use its content, including for text and data mining, without written consent. Contact: research@penmarkresearch.com.