MSFT — Microsoft Corporation

As of 2026-09-13 the sell-side is firmly constructive: 66 Buy vs 16 Hold and zero Sell ratings, with a consensus 12-month target around $553 (recent-print average $544.95, high $690, low $490) and the live brief citing firms nudging targets up on pullbacks. The tape supports it — +26.8% over three months and +25.3% over six — even though the last month was flat (+0.65%) and the last week slightly negative (-0.81%). The two live debates keeping sentiment from being euphoric are (1) whether the AI capex is converting: FY2026 free cash flow fell 12.3% YoY to $67.0B while capex rose ~80% to $115.9B, and (2) headline/legal risk from shareholder suits alleging Microsoft under-disclosed Copilot functionality and data-siloing issues. Both are being treated by the market as costs of a leadership position rather than a broken thesis.

Top line: structurally strong with a genuine re-acceleration. Quarterly revenue went from $56.5B (Sep-2023) to $90.0B (Jun-2026). YoY growth bottomed at ~12.3% in the Dec-2024 quarter and ~13.3% in Mar-2025, then re-accelerated to +18.4% (Sep-2025), +16.7% (Dec-2025), +18.3% (Mar-2026) and +17.8% (Jun-2026). Full FY2026 revenue of $331.8B was +17.8% vs FY2025's $281.7B, itself +14.9% vs FY2024 — so growth is not decelerating, it is stabilizing in the high-teens. Margins: operating margin expanded YoY in every quarter (Sep-2025 +230bps to 48.9%, Dec-2025 +160bps to 47.1%, Mar-2026 +60bps to 46.3%, Jun-2026 +20bps to 45.1%) and full-year FY2026 operating margin of 46.8% was ~116bps better than FY2025's 45.6%. But the sequence inside FY2026 is a clean 380bp sequential decline (48.9% → 45.1%), which lines up with depreciation and AI infrastructure costs ramping — that is the one genuinely cautionary trend in the dataset. Earnings: FY2026 net income $133.7B on EPS of $17.95 (sum of quarters), but note the Dec-2025 quarter is distorted — net income of $38.5B exceeded operating income of $38.3B and produced $5.16 EPS vs $3.72 the prior quarter and $4.27 the next, i.e. a large non-operating/one-off gain, not operating performance. Strip it and the underlying EPS cadence is closer to $3.72 → $4.27 → $4.80, a steadier low-teens-to-high-teens climb. Cash flow: this is where the story inverts. FY2026 operating cash flow was strong at $182.9B, but capex of $115.9B (fully 34.9% of revenue, vs 22.9% in FY2025) crushed free cash flow to $67.0B from $76.3B. Q4 FY2026 FCF of $19.6B was down 23% YoY, and the Dec-2025 quarter was an outright air pocket at $5.9B. Verdict: revenue and earnings trends are stable-to-accelerating and YoY margins are still expanding; the margin mix is softly compressing sequentially and FCF is structurally impaired by deliberate investment, not by demand weakness.

As of 2026-09-13, MSFT at $495.63 screens as fair, leaning modestly cheap on earnings and expensive on cash flow. On earnings: 27.5x TTM EPS of $18.00 and 25.1x FY2027 consensus EPS of $19.72, versus a 5-year average TTM P/E of 31.41 — that is roughly a 12% discount on trailing and ~20% on forward. PEG of 0.875 supports the 'not expensive' read. Two caveats. First, the trailing EPS is flattered: the Dec-2025 quarter included a one-off that pushed net income above operating income, worth roughly $1.20–$1.40 of EPS; normalized TTM EPS is closer to ~$16.6–$16.8, putting the real trailing multiple nearer 29.5–30x — still below the 5-year average, but not the bargain the headline number implies. Second, on cash the stock is not cheap at all: FY2026 FCF of $67.0B on a $3.68T market cap is a 1.8% free-cash-flow yield, and 18.3x EV/EBITDA and 11.09x sales (vs a 11.4x 5-year average P/S) leave no room for a growth wobble. The reason is mechanical — capex is running at ~40% of quarterly revenue. Bottom line: you are paying a below-history multiple for above-history growth, but you are paying a full multiple for cash flow that is temporarily suppressed by choice. Overall this is neutral-to-attractive, not stretched.

The stock is priced for the AI transition to work, but not priced for perfection — that is the key nuance. At 25x forward EPS, the market is already accepting roughly $390.6B of FY2027 revenue (+17.7% over FY2026's $331.8B) and $19.72 of EPS, and it is accepting the capex line as an investment rather than a cost. That means the multiple itself is not the risk; a mid-20s multiple is below MSFT's own 5-year average of 31.4x and roughly in line with a high-quality mega-cap grower. The risk is that earnings have to show up to justify even that: if Azure growth holds high-20s and Copilot seats convert, high-teens EPS compounding makes the current price look reasonable in 12 months. If capex keeps growing faster than revenue (as in FY2026, capex +80% vs revenue +18%), free cash flow stays pinned near $67B, the FCF yield stays under 2%, and the multiple grinds toward the low-20s — which is roughly a low-$400s stock on unchanged EPS. Watch the FY2027 capex guidance on Oct 28, 2026 more than the EPS print; that is the line that determines which of those two paths you are on.

As of 2026-09-13, the chart is a healthy uptrend with stalled short-term momentum. RSI(14) stood at 56.68 — neutral, neither overbought nor oversold. Price was +0.24% versus the 20-day SMA (about $494), +9.25% versus the 50-day SMA (about $454) and +14.93% versus the 200-day SMA (about $431), so the moving-average stack is fully bullish and the stock is not extended. The offset: return of -0.81% over one week and +0.65% over one month against +26.84% over three months means the advance has paused rather than continued. Entry-timing implication: this is a reasonable place to add on a long-term thesis because RSI leaves room and the trend is intact, but it is not a washed-out entry — the stock is ~9% above its 50-day. Better risk/reward sits on a pullback toward the $450–$460 zone (50-day SMA); a break below the 200-day near $431 would be the real trend-change signal to respect.

• Interest-rate expectations and long-duration multiple sensitivity: mega-cap software valuations are discount-rate sensitive, and rate stabilization has supported the current multiple but a re-acceleration of inflation would compress it quickly.
• The AI capital-expenditure cycle: GPU, networking and data-center power availability and pricing directly hit MSFT's depreciation and gross-margin line — capex was 34.9% of FY2026 revenue versus 22.9% in FY2025.
• Regulatory and antitrust scrutiny of Big Tech: cloud market dominance, AI ecosystem partnerships and software bundling remain live investigations in the US, EU and UK.
• Enterprise IT budget and cloud-migration spending: roughly half the business is tied to corporate capex cycles; a general enterprise-spend slowdown would hit Azure consumption and M365 seat expansion.
• Sector rotation and AI sentiment: MSFT trades as a proxy for the whole AI trade, so flows into or out of AI-themed equities move the stock independent of quarterly results.
• Securities-disclosure litigation environment: shareholder suits over Copilot disclosures keep headline risk elevated and can pressure management's willingness to give aggressive forward guidance.

• FQ1 2027 earnings on October 28, 2026 — the single most important event. Watch Azure growth rate, commercial M365 Copilot adoption/seats, operating-margin trajectory, and especially FY2027 capex guidance.
• Evidence that AI capex is converting into revenue: improving FCF trend (FY2026 FCF fell 12.3% YoY to $67.0B) or a stated peak in capex intensity would be the single biggest re-rating lever.
• Progress or resolution on the Copilot securities class action — removal of that overhang would let the multiple revert toward its 5-year average of 31.4x.
• Continued cost-management programs and capital returns (0.76% dividend yield plus buybacks) offsetting the capex drag.
• New AI product launches and ecosystem partnerships (Anthropic cloud commitments, NVIDIA hardware integrations) that expand the monetizable AI surface area.

• Capex outrunning returns: FY2026 capex rose ~80% to $115.9B while free cash flow fell 12.3% to $67.0B, and Q4 FY2026 FCF was down 23% YoY. A 1.8% FCF yield provides almost no valuation support if the AI payoff slips.
• Sequential margin compression: operating margin slid from 48.9% in the Sep-2025 quarter to 45.1% in Jun-2026, a 380bp deterioration that is only masked by easy year-ago comparables.
• Earnings-quality optics: the Dec-2025 quarter's $38.5B net income exceeded operating income, inflating TTM EPS to $18.00 — normalized trailing EPS is closer to ~$16.7, so the reported 27.5x P/E understates the true multiple.
• Legal and regulatory overhang: the Copilot disclosure class action, plus ongoing global antitrust scrutiny of cloud bundling and AI partnerships, creates both headline and potential remediation risk.
• Competitive pressure on AI share from Google's Gemini and rival models, which is precisely the share-loss allegation in the outstanding shareholder complaints.
• Valuation is full on cash metrics: 18.3x EV/EBITDA and 11.1x sales mean a growth disappointment compounds into multiple compression, not just an EPS miss.

Yes, this is a reasonable long-term holding at $495.63 as of 2026-09-13, with one honest caveat. The positives are real and improving: FY2026 revenue grew 17.8% (up from 14.9%), YoY operating margins expanded 116bps, ROIC is 20.6% and the balance sheet is a fortress at 0.29 debt/equity with a 1.23 current ratio — and you are paying 25.1x forward EPS against a 5-year average of 31.4x. The caveat is that the earnings-based cheapness is partly an artifact of a one-off gain inflating trailing EPS, and it absolutely does not extend to cash flow: FY2026 FCF fell to $67.0B for a 1.8% FCF yield because capex nearly doubled to $115.9B. So you are not buying a cheap stock; you are buying a high-quality compounder at a slight discount to its own history while underwriting a very large, very deliberate capital bet that hasn't yet shown up in free cash flow. If you believe AI monetization converts, the next two years look good; if capex keeps growing faster than revenue, the multiple has room to compress. Reasonable to accumulate, ideally on a pullback toward the 50-day SMA near $454 rather than chasing at $495.

Growth-adjusted forward P/E method: I applied a 28x multiple to the FY2027 consensus EPS of $19.724, giving $552.27 (28 x 19.724). The 28x comes from MSFT's 31.41x 5-year average TTM P/E discounted roughly 10% for the capex-driven FCF compression (FY2026 FCF -12.3% YoY) and the Copilot litigation/regulatory overhang, balanced against a 20.6% ROIC, 40.3% net margin and 17.8% revenue growth that justify a premium to the market. Cross-check: at the 2026-09-13 price of $495.63 the stock trades at 25.1x forward EPS, i.e. the market is already applying a ~20% discount to MSFT's own historical earnings multiple, so $552 implies only a partial reversion to that historical average, not a stretch.