Why S&P switched Samsung’s outlook to “positive”, and what it means for buyers, suppliers and investors
On July 21, 2025, S&P Global Ratings changed Samsung Electronics’ credit outlook to “positive” while reaffirming the issuer credit rating at AA- and the short‑term rating at A‑1+. The agency ties a potential upgrade to a scenario where AI-driven demand for server memory, especially DDR5 and high‑bandwidth memory (HBM), keeps prices high, supply tight, and Samsung’s cash generation materially stronger through 2026-2027.
S&P’s core assumptions and the headline numbers
S&P’s thesis ties three things together: (1) aggressive hyperscaler investment in AI data centers is raising server memory demand, (2) near‑term memory supply is constrained and supporting higher prices, and (3) if Samsung wins HBM and foundry share while maintaining yields, revenue, profitability and free cash flow could rise enough to make an upgrade plausible.
Key S&P figures (as reported by S&P):
- S&P reaffirmed Samsung’s long‑term issuer credit rating at AA- and short‑term rating at A‑1+, and revised the outlook to “positive” from “stable”.
- S&P projects annual revenue of about KRW 683 trillion in 2026 and KRW 821 trillion in 2027.
- S&P reported “earnings before depreciation and amortization” rising from around KRW 91 trillion in 2025 to KRW 393 trillion in 2026 and KRW 502 trillion in 2027, figures S&P presents as part of its upside scenario.
- S&P cites estimates that some DDR5 chip prices have “risen by three to four times year over year.”
- S&P expects memory‑market tightness to persist through 2026 and 2027 and notes that meaningful capacity additions from memory suppliers “may not materially arrive until after 2028.”
- The agency also referenced a projection that combined capex by the world’s four largest hyperscalers could climb to roughly $1 trillion in 2028, “around four times their 2024 level.” (S&P attributes this as a market assumption in its scenario work.)
“The ‘positive’ outlook is not unconditional.”, S&P Global Ratings
Two caveats. These are S&P projections and scenario assumptions, not guarantees. Several numeric claims, notably the quoted EBITDA‑like figures and the DDR5 price multiple, deserve close scrutiny because they look inconsistent with other published industry data or with simple arithmetic compared to the revenue projections. Readers should consult S&P’s full release for the precise metric definitions and underlying assumptions.
Technical and competitive threads S&P highlighted
- HBM execution: S&P says HBM3E yield issues “have been substantially improved” and highlights a possible HBM4 cost and packaging edge for Samsung via a 1c DRAM plus 4-nanometer base die approach. (HBM = stacked DRAM used on accelerators for high throughput; yields and packaging matter because HBM volumes are small but command a premium.)
- DDR5 strength: S&P points to sharp DDR5 price moves as evidence of tightening. DDR5 is the mainstream server DRAM standard and benefits as AI servers increase memory per node.
- Foundry opportunity: S&P flagged ongoing capacity constraints at TSMC as a potential window for Samsung Foundry, provided Samsung can deliver yields and customer qualifications at pace.
- Long‑term supply agreements (LTSAs): S&P notes multi‑year contracts (3-5 years) between suppliers and hyperscalers could reduce cyclicality and help stabilize cash flows.
Where S&P’s view is most fragile
S&P’s upside relies on three linked uncertainties: hyperscaler capex holding near the level S&P modeled, memory pricing staying elevated rather than collapsing as suppliers add capacity, and Samsung turning engineering gains into volume with healthy yields. Any of those can break the thesis:
- Hyperscaler behavior matters. S&P’s scenario references the “big four” hyperscalers (commonly AWS, Microsoft/Azure, Google/Alphabet and Meta). Their combined procurement cadence, inventory policies and architecture choices, for example memory per accelerator, will materially influence revenues and average selling prices.
- Supply responses can be swift. Memory fabs have long lead times, but suppliers can accelerate expansions or reallocate capacity. S&P notes meaningful new capacity may not arrive until after 2028, but that is an industry timing judgment, not a certainty.
- Metric consistency needs checking. The “earnings before depreciation and amortization” line S&P cites (KRW 91T → 393T → 502T) appears inconsistent with the revenue projections; this could reflect a unit or metric label issue in the S&P briefing. Treat those figures as S&P projections that require verification in the agency’s full note.
How this changes the playbook, concise actions for three audiences
- Hyperscalers and large buyers: Model higher memory costs into service economics and consider negotiating 3-5 year LTSAs to lock capacity and reduce spot-price exposure. Prioritize multi-sourcing where technical fit allows.
- Chip suppliers and fabs: Focus on yield improvement, packaging throughput (CoWoS/EMIB equivalents), and compressing customer qualification timelines. Execution will matter more than optimistic market forecasts.
- Investors: Treat S&P’s outlook as a directional signal that raises the probability of an upgrade under S&P’s scenario. But demand the earnings and cash-flow evidence: look for quarterly mix shifts toward memory and foundry and sustained ASPs before re-rating Samsung on fundamentals.
Top signals to monitor (priority checklist)
Watch these three leading indicators first. They will validate or undermine S&P’s scenario faster than broad market noise.
- Hyperscaler capex guidance and commentary. Quarterly signals from AWS, Microsoft, Google/Alphabet and Meta about AI-related builds and capex cadence are the primary drivers of server memory demand.
- DRAM/HBM contract ASP trends from market trackers. Monthly and quarterly reports from DRAMeXchange, TrendForce and similar services will show whether DDR5 and HBM ASPs are rising or rolling over.
- Samsung’s quarterly mix and yield commentary. Look for clear evidence of durable memory revenue growth, improving HBM3E yields, and disclosed foundry customer wins or ramp milestones.
Other useful trackers include announced memory supplier fab start dates and equipment order books, public long‑term supply agreements between cloud buyers and vendors, and any material acceleration in TSMC capacity announcements that would change the foundry competitive dynamics.
Short Q&A, likely questions executives are asking
- Why did S&P change Samsung’s outlook to “positive”?
S&P sees a plausible scenario where AI-led demand for DDR5 and HBM, near‑term supply constraints, and potential foundry share gains push Samsung’s revenue and cash flow materially higher through 2026-2027. That raises the probability of a ratings upgrade if those conditions hold. - Are the revenue and profit numbers S&P cited guarantees?
No. The figures (for example, KRW 683 trillion revenue in 2026 and KRW 821 trillion in 2027, and the quoted earnings‑before‑depreciation‑and‑amortization series) are S&P projections and scenario assumptions. They depend on pricing, mix and execution and should be checked against S&P’s full note for metric definitions. - How much depends on hyperscaler behavior?
A lot. S&P’s upside rests on assumptions about combined hyperscaler capex and purchase behavior. S&P cited a projection of roughly $1 trillion combined capex by the four largest hyperscalers in 2028. Those buyers’ procurement strategies will heavily influence memory pricing and supplier revenue. - What could reverse the outlook?
A sharp pullback in memory demand, faster-than-expected capacity additions by suppliers, or renewed technology or yield setbacks at Samsung would all reduce the odds of an upgrade and could revert the outlook to “stable.”
Final read for executives
S&P’s move to a “positive” outlook is directional: it raises the probability that Samsung could be upgraded if AI demand, memory pricing and Samsung’s execution align. Many moving parts remain, hyperscaler capex, supplier capacity timing, and yield and qualification execution. For buyers, prioritize supply stability and long‑dated agreements; for suppliers, prioritize execution; for investors, wait for quarter-to-quarter financial proof points. Numbers, not narratives, will determine whether this hopeful outlook becomes reality.