CG Power — What Changed Since June (and Whether to Buy the Fall)

CG Power — What Changed Since June (and Whether to Buy the Fall)

Update date: 19 October 2026 | CMP: ₹854 | Model: deepseek-v4-flash (DeepSeek) | Previous analysis: 17 September 2026 at ₹882

This began as a follow-up to our 18 June 2026 deep dive and is now on its second refresh. It covers what has changed in the business, prices the stock forward, and — because that is what was asked — answers directly whether this is a good investment while the market is falling.

---

0. 19 October Update — What Changed in a Month, and the Short Answer

This section is new. Everything below it was written on 17 September 2026 and is kept as the record of that day's view; where a number has moved, this section supersedes it.

The stock: ₹854 on 19 October against ₹882 a month ago — -3.2%, and -12.9% from the ₹981 all-time high. Over the same month the Nifty fell 2.1%; over six months it is down 6.6%. So CG Power has tracked the index on the way down, and today it is outperforming: +0.95% while the Nifty is -1.28%.

The tape inside that modest net move has been violent: ₹863 (14 Oct) → ₹818 (15 Oct, -5.2%) → ₹846 (16 Oct, +3.4%) → ₹854 today. Delivery on those three sessions was 45.5%, 47.6% and 50.8% — rising into the bounce, which reads as accumulation rather than distribution.

Three business facts changed:

1. Q2 FY27 results are on 23 October — board meeting, with the earnings call the same day and senior management including the MD, CFO and segment heads. Consensus needs EPS of about ₹2.66 after the Q1 miss at ₹1.99. This is the largest swing factor in the next three months.
2. G2's timeline moved out. Manufacturing from the second OSAT unit is now guided to Q4 CY2027 or early 2028, with construction and tool move-in by Q2 CY2027. The "2027 re-rating catalyst" our September note carried is a 2028 revenue event, not a 2027 one.
3. The semiconductor order backlog is expected to double if execution holds, with Renesas committed to roughly 50% of combined G1 and G2 capacity and additional customers in discussion. The design arm (Exero/Axiro) is revenue-generating and expanding from RF and SATCOM into power electronics.

Two numbers got worse. Working capital days rose from 31.3 to 69.8, and the stock still trades at 105-108x trailing earnings and 16.7x book. A falling market fixes neither of those.

Is it good to invest in it in a down market? The full answer is in section 8.4. In one line: the business is good, the price is not cheap, and the next four days carry event risk — so accumulate in tranches around the 200-day moving average, not in one go at ₹854.

---

1. Then vs Now — What Actually Changed

The headline: the business kept performing, the stock de-rated modestly, and the one thing our June note could not do — value the stock forward — is now possible, because the market has finally set consensus estimates for FY27 and FY28.

---

2. What Happened Since 18 June

2.1 Q1 FY27 results (reported 27 July 2026) — strong operationally, one miss on the bottom line

Standalone (core business, excluding subsidiaries): sales ₹3,061 Cr (+16%), PAT ₹364 Cr (+27%), PAT margin 11.9% vs 10.8%, operating margin +140 bps, ROCE 23% for the quarter.

Segment detail — the divergence is the story:

Power Systems backlog hit ₹14,434 Cr (+59% YoY). Industrial Systems was dragged by a ₹20 Cr one-off provision in the railways business and weak consolidated industrial margins (7.6% vs 10.2% YoY) — management attributed the gap to G.G. Tronics, which is still pre-revenue and "at breakeven today."

The miss: consensus FY27 Q1 EPS was ₹2.20; actual was ₹1.99 — a -9.7% surprise, the first negative earnings surprise in four quarters (Q3 FY26 +3.4%, Q4 FY26 +12.9%). The shortfall came from lower Industrial Systems margins and higher semiconductor losses, not from the power business.

Other operational highlights from the call:

• Order intake ₹4,692 Cr standalone / ₹5,211 Cr consolidated in the quarter
• Transformer capacity now ~75,000 MVA (power) plus ~10,000 MVA (distribution), up from 22,000 MVA four quarters earlier; the greenfield plant ramps 10,000 → 30,000 → 45,000 MVA across Q1-Q3
• Price hikes of 17.5% plus a further ~5% taken to offset raw material inflation — "the market is absorbing it pretty well"
• Export order intake roughly doubled (+84% YoY); management targets a spread of geographies (utility, renewable, data centre, oil & gas) rather than dependence on any one
• Motors delivered double-digit growth for a fourth consecutive quarter — the turnaround is now a trend, not a quarter
• Cash remains fortress-like: interest cost for the quarter was just ₹4 Cr on ₹3,281 Cr of sales

2.2 Statutory auditor change (disclosed 24 July, effective 14 August 2026)

SRBC & Co LLP (EY network) resigned as statutory auditor effective 14 August 2026, to align with the audit firm of the new holding company — S.R. Batliboi & Co (also EY network). Both firms are EY affiliates, and no audit dispute or qualification has been disclosed.

Why this matters: the headline word "resignation" is a red flag in Indian markets for good reason, and the stock's volatile reaction in the following weeks reflects that nervousness. But the substance — an EY-firm-to-EY-firm switch to align with the parent — is procedural, not a governance event. We flag it because it is a legitimate reason sentiment wobbled, and because shareholders should note it rather than be surprised by it. Watch for confirmation in the FY27 audit.

2.3 CG Semi: G1 OSAT moves from "commissioned" to "commercially producing"

Since June, CG Semi announced commencement of commercial production at its G1 OSAT facility in Sanand, Gujarat, and reiterated a ₹7,600 Cr investment plan over five years across the two facilities (G1 and G2), built with Renesas (Japan) and Stars Microelectronics (Thailand). The fabless design arm, Exero Semiconductor (acquired from Renesas for ₹284 Cr), is already revenue generating, and management indicated Renesas has committed to offtake a substantial share (~40-50%, per analyst questioning) of output.

This is the transition our June note flagged as the potential 2027 re-rating catalyst: the semiconductor business has moved from construction to revenue, even though it is still loss-making.

2.4 Price action: an all-time high, then a two-and-a-half-month consolidation

• 1-2 July 2026: all-time high ₹981.15, immediately followed by a -5.5% intraday drop — the stock's steepest single-day fall in 15 months, on below-average volume (profit-booking, no confirmed trigger)
• July-November drift: the stock has traded a broad ₹858-930 range since, including a -5.6% day on 15 September (close ₹858) that took it to the bottom of that range
• Now: ₹882 on 17 September, -10% from the all-time high, and -7.5% below our June analysis price
• (19 October: ₹854, -12.9% from the high, after a ₹818 low on 15 October — see section 0)

---

3. Old Thesis vs New Reality — Marking Our Own Homework

Honest summary: the business read was good, the entry-price discipline was good, and the valuation framework was too crude — a trailing PEG on a company in a margin-expansion phase understated the forward growth. That is exactly the gap this update fixes.

---

4. Forward Valuation — The Section We Owe You

The multiples in this section were computed at ₹882 on 17 September 2026 and are left as written. At ₹854 they are about 3% lower — roughly 103x trailing rather than 106x. The forward argument does not change with a 3% price move; only the earnings print does.

4.1 Consensus estimates (20 analysts, September 2026)

Estimate revision trend: FY27 EPS consensus has been cut from ₹10.36 (60 days ago) to ₹10.08 — the Q1 miss is priced in. FY28 has held at ~₹13.93 throughout.

4.2 Forward multiples at ₹882

The clearest way to see it: the company's own revenue guidance trajectory (25% CAGR FY26-28 modelled by Motilal Oswal, with EBIT margins of 13.9% FY27 and 15.1% FY28) takes EPS from ₹7.86 to ₹13.93 in two years. On that path, the forward PEG is ~1.7 — expensive but defensible, versus the 11.3 trailing PEG that made the stock look indefensible in June. The valuation case is entirely a function of whether you believe the FY28 numbers.

4.3 Scenario table (FY28 horizon, at ₹882)

Even the bear case leaves the stock at 82x forward earnings — which is the honest cost of owning a best-in-class franchise with a semiconductor option embedded. There is no scenario in which this becomes a "cheap" stock on FY28 numbers; the question is only whether growth keeps compounding past FY28.

4.4 Broker targets — and what they imply

Note the shape of this: the average Street target (₹934) is only ~6% above the current price, and the low target (₹627) implies a 29% drawdown. The high target implies 25% upside. That is an unusually wide dispersion for a ₹1.4 lakh crore company, and it is entirely driven by how each analyst values the semiconductor business.

4.5 Credibility check — what has to be true

• Consensus FY28 EPS of ₹13.93 implies PAT of ~₹2,190 Cr — a 35% CAGR from FY26's ₹1,199 Cr, from a company that just missed quarterly estimates by 9.7%
• That requires: Power Systems holding 22-23% PBIT margins at roughly double today's revenue; Industrial Systems recovering from 8.9% back toward 12%; and semiconductor losses shrinking rather than widening
• It also implicitly assumes ~20-25% of FY28 profit comes from businesses that barely exist in the P&L today (OSAT G2 at full ramp)
• To justify today's ₹882 on a more sober 40x exit multiple, EPS needs to reach ~₹22 — implying roughly FY29-30 on the consensus trajectory. In other words, you are buying three to four years of compounding in advance, with the semiconductor option paying for the premium if it works

Verdict on valuation: not the "absurd" story the trailing PEG suggested in June, but not comfortable either. It is a high-quality compounder priced for its bull case in the nearer years and its base case beyond FY28.

---

5. What Hasn't Changed

Being honest about the parts of our June bear case that survive intact:

• Industrial Systems remains the weak engine — +6% sales, PBIT down to ₹148 Cr, consolidated industrial margin 7.6% vs 10.2%. G.G. Tronics is still pre-revenue. This has not improved in three months.
• Semiconductor losses still drag the consolidated P&L. G1 is now producing commercially, but the business is loss-making and G2 is still to come. Every quarter of drag is a quarter where consolidated growth (16%) lags standalone growth (27%).
• Other income is doing real work — ₹84 Cr in Q1 FY27 against PBT of ₹423 Cr (20%). Strip it out and the core P/E is higher still. It is legitimate income from the QIP corpus, but it is not operational.
• FIIs are still net sellers — 12.69% (Jun'25) → 11.97% (Jun'26), a seventh consecutive quarter of decline. DIIs (14.22% → 18.23%) are absorbing the supply, which is constructive, but the foreign bid has not returned.
• Execution load is unchanged and heavy — 1,10,000 MVA transformer target, switchgear greenfield, OSAT G1 ramp plus G2 build, railway approvals. Four simultaneous large programmes.
• The absolute valuation is still high on every historical metric: 109x trailing, 63x FY28, 45x FY28 EV/EBITDA, 1.3% earnings yield.

---

6. Shareholding — Steady Institutional Rotation

The pattern is unchanged and important: domestic institutions keep buying, foreign investors keep trimming, promoters stay put at 56.36%. DIIs have added 4 percentage points in a year. This is the strongest single argument that institutional conviction in the India power-capex story has not faded — it has just changed hands, from foreign to domestic.

---

7. Technical Analysis and 25-Point Checklist

7.1 Indicator panel (19 October 2026, CMP ₹854)

Structure: the consolidation has widened to roughly ₹818-935 and the stock is now in the lower half of it, having made a marginal lower high near ₹898 in October. Price sits below the 20/50/100-day cluster (₹872-897) but comfortably above the 200-DMA at ₹804 — the classic "pullback inside an uptrend" shape, not yet a trend break.

Interpretation: the 15 October drop to ₹818 was the second -5% day in a month, and it came with delivery at 47.6% — real transfers, not a volume vacuum. The two sessions since have recovered half of it on rising delivery (50.8% on 16 October) and above-average volume. Combine that with today's relative strength (+0.95% against a Nifty down 1.28%) and the tape is more constructive than the -12.9% from the high suggests. What it still lacks is a close back above ₹880-900, which is where the averages and the October high converge.

7.2 All 25 points assessed

Score: 18.5 / 25 (one item N/A) — down from 19.5 in September, and the change is almost entirely about trend and timing rather than the business: the 200-DMA is holding, delivery is rising and relative strength is genuinely good, but price sits below all three shorter averages and the market itself is falling. On the checklist's own rules this is not an entry signal yet — it needs either the 23 October print to clear and a close above ₹880-900, or a controlled retest of the 200-DMA that holds.

---

8. Updated Levels and Verdict

8.1 Levels

8.2 Rating

June 2026: "Expensive, quality business — not a value buy. Wait for a dip to ₹800-850." September 2026: "Accumulate on dips — quality compounder, priced for its base case and beyond." October 2026: Accumulate in tranches — but the price is not the signal, the 23 October print is.

The business case has strengthened since September, not weakened: order book ₹18,965 Cr (+45% YoY), Power Systems backlog ₹14,434 Cr with 23% segment PBIT margin, delivery rising to 50.8%, and one-year relative performance of +12.3% against a Nifty that fell 11.1%. The stock has de-rated 3.2% over that month while the order book grew.

What has got worse is everything about timing and price: the stock is below all three shorter averages, the market is in a downtrend, G2 revenue slipped to 2028, working capital days doubled to 69.8, and it still trades at 105-108x earnings and 16.7x book. In September the average Street target implied ~6% upside; from ₹854 it implies more, but only if the earnings hold.

It stays "accumulate", not "buy". A 105x multiple with a 0.95% earnings yield is a growth stock priced on perfection; in a falling market that is a liability, not a moat.

8.3 What I would do

8.4 Is this a good investment in a falling market?

This is the question this update was commissioned to answer, so it gets a direct one.

What a falling market does to a stock like this. CG Power trades at roughly 105x trailing earnings — an earnings yield near 0.95%. When the benchmark is down 5.6% over three months and long yields are rising, the first thing sold is the most expensive growth. High-multiple capital goods de-rate twice: once with the market, and again as the discount rate on their future earnings rises. A 13% drawdown from the high is not evidence that the de-rating has finished — at 16.7x book, a 20% drawdown from here would still leave it expensive.

What is genuinely good:

• Debt-free balance sheet — interest cost of ₹4 Cr a quarter on ₹3,281 Cr of sales
• ROCE 26.7%, ROE 20.5%, and an order book of ₹18,965 Cr, up 45% year-on-year
• Power Systems backlog ₹14,434 Cr (+59% YoY) at a 23% PBIT margin — the engine is working
• DIIs have added 4 percentage points in a year (18.23%) while FIIs trimmed to 11.97% — domestic money is absorbing foreign supply
• Delivery at 50.8% on the latest session, against a ~37% large-cap baseline
• Relative strength: +12.3% over a year when the Nifty fell 11.1%, and outperforming again today

What is genuinely risky:

• 23 October. A second consecutive EPS miss, in this market, likely breaks ₹818 and tests the 200-DMA at ₹804. Consensus wants ₹2.66 against ₹1.99 last quarter.
• The semiconductor business is still a cost. G2 revenue is now a 2028 event. Until then the OSAT operation subtracts from consolidated margin — ₹43 Cr, or 132 bps, in Q1.
• Industrial Systems remains the weak engine — PBIT ₹148 Cr against ₹172 Cr, with G.G. Tronics not yet contributing.
• Working capital days doubled to 69.8. In a tightening-rate environment that is a real cost, not a footnote.
• Valuation offers no protection. There is no dividend (0.15% yield) and no asset backing (16.7x book) to catch a fall.

The answer:

Position sizing: cap at 3-4% of portfolio, across at least three tranches, sized so that a gap through ₹786 on results costs a normal single-trade loss. The semiconductor optionality is a 2028 binary; it should not become a concentrated bet.

---

9. Catalysts to Watch Next

---

10. Disclaimer & Sources

This analysis is for information and education only. It is not investment advice and not a recommendation to buy, sell or hold. It was prepared by an AI agent (deepseek-v4-flash via Hermes Agent) from public sources. Consensus estimates are third-party forecasts and are frequently wrong — the FY28 numbers this valuation rests on are forecasts, not facts. Verify all figures against company filings before acting.

Sources

• TradingView scanner for NSE:CGPOWER (19 October 2026, intraday) — price, RSI(14) 46.2, SMA 20/50/100/200 at ₹872/₹881/₹897/₹804, ATR ₹28, 52-week range, relative performance
• NSE bhavcopy (sec_bhavdata_full) for 14, 15 and 16 October 2026 — closes of ₹863, ₹818 and ₹846, and delivery percentages of 45.5%, 47.6% and 50.8%
• NSE corporate filing and the company's earnings-call notice — Q2 FY27 board meeting and call on 23 October 2026
• Economic Times (Oct 2026) — CG Semi's second OSAT unit: construction and tool move-in by Q2 CY2027, manufacturing from Q4 CY2027 or early 2028
• Financial Express (Oct 2026) — semiconductor backlog expected to double; Renesas committed to ~50% of combined G1/G2 capacity; ISM 2.0 guidelines under review; working capital days 31.3 → 69.8
• Screener.in consolidated page for CGPOWER (19 Oct 2026) — P/E 105-108x, book value ₹50.6 (16.7x book), ROCE 26.7%, ROE 20.5%, dividend yield 0.15%, debt-free
• TradingView index scan (19 Oct 2026) — Nifty 22,850, -1.28% on the day, -5.6% over three months, -6.6% over six
• Our own 18 June 2026 analysis (stocksfundamentals.online/analysis/cg-power-deep-dive-jun18) — basis for the then-vs-now comparison
• Screener.in consolidated page for CGPOWER (logged-in, 27 sidebar metrics verified, 17 Sep 2026) — current P/E 109x, ROCE 26.7%, mcap ₹1,38,931 Cr, quarterly results Jun 2023-Jun 2026, shareholding through Jun 2026
• CG Power Q1 FY27 earnings call (reported 27 July 2026; transcript via Investing.com, Quartr, AlphaSpread) — segment PBIT, order intake and backlog, capacity ramp, management commentary on price hikes, G.G. Tronics and OSAT
• Yahoo Finance analyst estimates page for CGPOWER.NS — FY27/FY28 revenue and EPS consensus (20 analysts), estimate revision history, earnings surprise record
• Investing.com consensus page — 20-analyst average target ₹934.15 (low ₹627, high ₹1,100); Motilal Oswal Buy ₹1,020 (3 Sep 2026)
• Motilal Oswal research (May 2026) — 25% revenue CAGR FY26-28E, EBIT margins 13.9%/15.1% FY27/FY28, 32% PAT CAGR
• NSE corporate filing (24 July 2026) — SRBC & Co LLP intent to resign effective 14 August 2026, alignment with holding company auditor S.R. Batliboi
• Stocktwits / CG Semi disclosures — ₹7,600 Cr OSAT investment across G1 and G2, Renesas and Stars Microelectronics partnerships, Morgan Stanley Overweight with ₹799 base / ₹1,044 bull case
• Univest — CG Power's 5.5% single-day fall after the ₹981.15 all-time high (steepest in 15 months)
• NSE bhavcopy (2-15 September 2026) — delivery percentages and volumes
• Yahoo Finance price history — 251 sessions used for RSI(14), SMA 20/50/100/200 and 52-week levels

Data limitations: consensus estimates are for FY27/FY28 and carry wide dispersion (FY28 EPS range ₹10.74-₹19.10), so the forward valuation is a range, not a point estimate. The ₹3,000 Cr QIP corpus is treated as cash for EV purposes although a portion has been spent on capex. Semiconductor segment detail is deliberately opaque in the company's disclosures — neither the revenue run-rate nor the quarterly loss is broken out clearly, which limits the accuracy of any sum-of-the-parts valuation.