SYNTHOS RESEARCH

Navitas Semiconductor NVTS

Technology · Semiconductors · Synthos Deep Dive · 2026-07-19

$11.46
Avoid

The Overview

Navitas makes advanced power-conversion chips out of two newer materials — gallium nitride (GaN) and silicon carbide (SiC) — that waste less energy than ordinary silicon when converting electricity. That matters a lot for AI datacenters: NVIDIA's next-generation racks will draw so much power (up to a megawatt each) that the industry is redesigning how electricity gets into them, moving to an 800-volt architecture. NVIDIA publicly named Navitas as one of the chipmakers helping build that — and the stock exploded on the news, from about $5-6 to as high as $34.

Here is the problem: being named is not the same as being paid. NVIDIA named at least 14 chip companies to the same program, including giants like Infineon and Texas Instruments. Nobody — not NVIDIA, not Navitas, not any analyst we can find — has said how many dollars of Navitas chips go into each rack, or whether Navitas wins any socket at all. Meanwhile the actual business shrank: sales nearly halved last year to $46M, the company currently loses money on every chip it sells (negative gross margin in two of the last three quarters), it burns roughly $25M a quarter running the company, and it has paid for that by issuing new shares — 53% more shares exist now than in 2022. Wall Street's own forecasts say 2026 sales will be lower than 2025. And company insiders, including the CEO and a director, sold millions of dollars of stock near the top in May 2026 (SEC Form 4 filings).

Our three scores in everyday terms:

The one big thing to remember: the stock already did the round trip — $5.44 to $34.17 and back to $11.46 inside a year. The market has voted twice on this story. Until sales actually inflect, price is attention, not value.


Putting a number on it: our fair-value estimate is $7.50 against a current price of $11.46 — consistent with our call to stay away or wait for a better setup.

Our summary metrics

Downside Risk (lower = safer)
8/10 · Very High
Negative gross margin in 2 of the last 3 quarters, ~$25-27M operating loss per quarter, beta 3.8, +53% share dilution since 2022, heavy insider selling at $28-32 in May 2026, and a −64% drawdown already on the books. The only mitigant — ~$230M net cash (2025-12-31) funding roughly 4-5 years of burn at the FY25 pace — keeps this an 8, not higher.
Growth Quality
1/10 · Low
Revenue fell −44.9% in 2025 ($83.3M → $45.9M) and consensus models 2026 at ~$42.4M — BELOW 2025. Gross profit was negative in Q3-25 (−$0.9M) and Q1-26 (−$1.5M). There is no reported growth to grade quality on; every growth claim is forward-looking hope, not evidence. A 1.
Exponential Potential
5/10 · Moderate
The optionality is genuine — a named NVIDIA 800 V HVDC silicon partner with both GaN and SiC, and consensus hockey-sticks to $500M by 2030E — but Navitas is one of 14+ named partners, no source quantifies its dollar content or share, the flagship demos are boards not products, and China (Innoscience) is undercutting trailing-edge power silicon. Real option, unpriceable odds — a 5.
Fair value$7.5 $5–$16
What does “fair value” mean?

Fair value is Synthos’s estimate of what one share is worth today, based on our model of the company’s future cash generation and the risks to it. It is not a price target or a prediction of where the stock trades next quarter — it’s the price where we think risk and reward are balanced. Above it, you’re paying for outcomes better than our base case; below it, the market is offering a margin of safety.

The model’s inputs and weightings are proprietary — they’re the product. What we publish is the output, We don’t publish a reverse-DCF cross-check for pre-profit companies — negative or missing earnings break that math — so take this number on our modeling alone.

What do the 5 tiers mean? (Core · Tactical · Watch · Hold · Avoid)
Buy — CoreOwn it as a foundation — start or add now, size it for years, let dips be gifts.
Buy — TacticalGood price + confirmed trend + a defined exit — buy the setup, not a marriage.
WatchWe want the business, just not at this price/setup — act only when the listed trigger hits.
HoldFine to keep if you own it — no reason to buy more; new money does better elsewhere.
AvoidDon't own it — the problem is the business or the expectations, so a cheaper price won't fix it.

The Road Ahead

What we expect to matter in each window, and the evidence that would prove us wrong.

Short term 0–6 months

No differentiated view
Driver
The story-stock round trip is in progress, not finished — $11.46 sits −64% off the EOD high, below both the 50-DMA ($20.59) and 200-DMA ($12.66), RSI 14 at 24, with a Q2 print (~2026-07-27) that consensus expects at ~$10.0M revenue and another loss, and Kyber-slip headlines (SemiAnalysis via CNBC, 2026-07-06) hanging over the theme.
What we’re watching
A Q2 print with positive gross margin and any quantified data-center/800V revenue would firm the floor; another negative-gross-margin quarter or a fresh equity raise would open the $5.44 attention-floor zone.
Confidence
Medium

Medium term 6–24 months

No differentiated view
Driver
Meaningful 800 V revenue physically cannot arrive before the NVIDIA Kyber rack ramp — officially 2027, reported by SemiAnalysis (via CNBC, 2026-07-06) as slipping to 2028 on midplane yields, denied by NVIDIA — while consensus has Navitas at just ~$72.9M (2027E) and ~$124.6M (2028E) with EBITDA still negative through 2029E. The stock must live on announcements for at least four more quarters.
What we’re watching
A committed-volume design win with disclosed dollar content, two consecutive quarters of positive and rising gross margin, or Kyber 2027 confirmation would each be genuine flip evidence; estimate cuts or ATM issuance into spikes deepen the case against.
Confidence
Medium

Long term 2+ years

Tailwind
Driver
The 800 VDC datacenter power transition is a real architectural shift (1 MW racks make 54 V distribution untenable), and wide-bandgap GaN/SiC content per rack should grow for a decade — a legitimate secular tailwind for whoever wins the sockets.
What we’re watching
The tailwind is to the category, not automatically to Navitas — 14+ named NVIDIA silicon partners including Infineon, TI, onsemi and ST, plus China GaN volume leader Innoscience undercutting on price, mean Navitas must convert demos into quantified share to participate. Watch for disclosed per-rack dollar content and named socket wins.
Confidence
Low

Exponential Potential

Exponential Potential
5/10 · Moderate
The optionality is genuine — a named NVIDIA 800 V HVDC silicon partner with both GaN and SiC, and consensus hockey-sticks to $500M by 2030E — but Navitas is one of 14+ named partners, no source quantifies its dollar content or share, the flagship demos are boards not products, and China (Innoscience) is undercutting trailing-edge power silicon. Real option, unpriceable odds — a 5.

What could take this further than the base case — and where the market may be underpricing it. Full forward-growth and acceleration math in Deeper analysis, §4 below.

Deeper analysis

Technicals, fundamentals, valuation, and the full expert-claim evidence panel — the detail behind the numbers above.


Reference table

Street consensusPT $18.23 (high $21 / low $13.70) but ratings say Hold: 0 Strong Buy · 3 Buy · 4 Hold · 1 Sell — targets look stale from the spring spike
Valuation~69× TTM sales (P/S) · ~64× TTM EV/S · ~61× 2026E EV/S · ~75-81× annualized Q1-26 sales · P/B 6.1× · no earnings to multiply (TTM EPS −$0.60)
TechnicalsBroken — $11.46, −64% off the 52-wk high ($31.79, EOD basis), below the 50-DMA ($20.59) and 200-DMA ($12.66), RSI 24, MACD −2.56, still +86% 12-mo (SPY ~+19%)
ConvictionNone — 0 net-bullish voices, 0 traceable KB claims. Call rests on fundamentals + quant only
Position sizing0% — Avoid. A rules-based parabola/attention trade is a different instrument (§12), not an investment allocation
Single biggest risk (to a holder)The valuation needs the 800 V story to convert to revenue on schedule; the revenue physically cannot arrive before the Kyber ramp (2027 target, reported 2028 slip), and Navitas's share of it is unquantified anywhere

What the experts actually said

No independent expert claims in the Synthos knowledge base yet for NVTS — this dive is fundamentals- and technicals-driven, not panel-driven.

Price & moving averages 12 months · 50 & 200-day averages · 52-week range

19182634Jul '25Sep '25Dec '25Feb '26May '26Jul '2652w hi $3250-DMA 21200-DMA 13Price 1152w lo $6

Solid line = price · dashed line = 50-day average · dotted line = 200-day average · the two thin horizontal lines mark the 52-week high and low. Price above both averages is an uptrend.

Data summary: last close $11.46, 44% below the 50-day average ($21), 9% below the 200-day average ($13) — a downtrend. 64% below the 52-week high of $32, 107% above the 52-week low of $6.

Bollinger Bands 20-day average ± 2 standard deviations

09182736Jul '25Sep '25Dec '25Feb '26May '26Jul '2620-day avg 16Price 11

The shaded band widens when the stock gets more volatile. Riding the upper edge = strong momentum (sometimes stretched); the lower edge = weak / potentially oversold.

Data summary: price $11.46 is currently inside the band (band $9–$23).

RSI (14) momentum gauge · 0–100

705030Jul '25Sep '25Dec '25Feb '26May '26Jul '26RSI 30.9

Above 70 (overbought zone, shaded) = overbought, below 30 (oversold zone, shaded) = oversold. Currently 31.

MACD 12 / 26 / 9 · trend & momentum

0Jul '25Sep '25Dec '25Feb '26May '26Jul '26signal -2.2MACD -2.6

The MACD line crossing above the signal line (bars flip to the up color) = momentum turning up; crossing below (bars flip to the down color) = turning down. Bar height = the size of that gap.

Data summary: MACD is currently below its signal line by 0.31, negative momentum.

Relative performance vs S&P 500 & its sector (XLK (sector)), set to 100 a year ago

55176298419541Jul '25Sep '25Dec '25Feb '26May '26Jul '26NVTS 183XLK (sector) 135S&P 500 118

Solid = NVTS · dashed = S&P 500 · dotted = XLK (sector). A rising line means it is beating that benchmark — the sector line shows whether it is a leader or laggard within its own group.

Forward revenue & earnings actual → estimate · "FY" = fiscal year, "E" = estimate

00001$0BFY23EPS $-0$0BFY24EPS $-0$0BFY25EPS $-0$0BFY26EEPS $-0$0BFY27EEPS $-0$0BFY28EEPS $-0$0BFY29EEPS $-0$0BFY30EEPS $0

Darker bars = actual results, brighter = analyst estimates. Taller bars to the right = expected growth.

Key stats an RIA wants

Price$11.46
Market cap$3B
P/E trailingn/m (loss-making or n/a)
P/E FY26E / FY27En/m (loss-making or n/a) / n/m (loss-making or n/a)
EV / Sales63.6×
EV / EBITDA-23.3×
Gross margin6.7%
Net margin-330.7%
Dividend yield0.00%
Beta3.815
52-wk range$6 – $32
RSI(14)24
50 / 200-DMA$21 / $13
12-mo return+86% (SPY +19%)
Street target$18 ($14–$21)
Analyst grades3 Buy · 4 Hold · 1 Sell
FMP ratingC
Next earnings~2026-07-27 (Q2 2026 earnings; watch gross margin sign, cash burn, any data-center/800V revenue disclosure, and share count)

1. What it is

Navitas Semiconductor (Nasdaq: NVTS) designs and sells wide-bandgap power semiconductors: GaNFast gallium-nitride power ICs, GeneSiC silicon-carbide devices, plus silicon system controllers and isolators, for power conversion and charging (FMP profile). End markets span mobile/consumer fast chargers (its original volume business), solar, EV/automotive, and — the entirety of the current story — AI datacenter power. Founded 2014, Torrance CA, ~190 employees, CEO Chris Allexandre; came public via SPAC on 2021-10-20. It is fabless and tiny relative to the power-semi incumbents (Infineon, onsemi, ST, TI).

The story that moved the stock: on May 21, 2025 NVIDIA selected Navitas (8-K filed; EDGAR CIK 1821769) to co-develop the 800 V HVDC power architecture supporting Kyber rack-scale systems and Rubin Ultra — 1 MW IT racks — using both GaNFast and GeneSiC. Navitas was honored at NVIDIA's Partner Ceremony in Taipei (May 29, 2026), and at GTC 2026 (March) and Computex 2026 demonstrated a single-stage 800V→6V GaN power-delivery board (16× 650V GaNFast FETs) that eliminates the 48V intermediate bus, with claimed peak efficiency of 96.5% (March) rising to 97.5% (June) at 1 MHz and 2,100 W/in³ power density. Those are demo-board targets, not shipping product, and not third-party validated. The essential context: NVIDIA's 800 VDC partner list names at least 14 silicon companies — Infineon, TI, onsemi, ST, MPS, ROHM, ADI, Renesas, Innoscience, AOS, EPC, Power Integrations, Richtek, and Navitas — so being named is table stakes for the category, not a moat. (All facts this paragraph: Synthos verified-research run, 2026-07-19, sources as cited.)

Revenue mix: FMP segment data shows a single reportable segment ($83.3M, FY2024 — the last year broken out); the company does not disclose a datacenter/800V revenue line at all, which is itself a tell. Geography (per profile): US, Europe, China, rest of Asia — with China both a major end market and, via Innoscience, its most dangerous competitor.

2. The expert thesis — why the panel is bullish (traceable)

There is no expert coverage of Navitas in the Synthos knowledge base. total_claims = 0, net_bullish_voices = 0. No tracked voice in our panel has published a distilled, traceable claim on this name.

That means this deep dive carries no conviction rating and cites zero claim_ids — because there are none, and fabricating conviction is against the house standard. The verdict below is entirely fundamentals- and quant-driven: reported financials, the analyst-consensus estimate path (labeled as estimates), the adversarially-verified public record on the NVIDIA program, and Synthos's own scoring. The one piece of tracked-voice context worth flagging cuts against the name, not for it: Dylan Patel (SemiAnalysis; KB claims dated 2026-02-03/05) argues China power-chip makers are "matching and undercutting TI/ST Micro" in power semiconductors — his thesis names TI and ST, not Navitas, so we cite it only as thematic context: power semis are trailing-edge silicon where China (and specifically Innoscience, the China GaN volume leader) competes on price. Readers who require expert-panel corroboration should treat this as a quant/fundamental screen, not a conviction call.

3. Synthos scores & the Bull / Base / Bear cases

The one-glance judgment — three scores, 0–10, each anchored to real metrics (not probabilities we can't honestly calibrate):

Score0–10The read
Downside Risk (lower = safer)8 · HighGross profit negative in Q3-25 (−$0.9M) and Q1-26 (−$1.5M); operating loss ~$25-27M/quarter (Q4-25 −$24.8M, Q1-26 −$27.3M); beta 3.8; diluted shares 145.7M (2022) → 222.3M (Q1-26), +53%, including $202.5M of stock issued in FY2025 alone (cash-flow statement) — sold into the NVIDIA-story strength; CEO and directors sold at $28-32 in late May 2026 (Form 4s, SEC). The mitigant: ~$230M net cash (cash $236.9M vs $6.5M debt, 2025-12-31), roughly 4-5 years of runway at the FY25 burn (OCF −$42.9M). Solvency is not the risk; permanent capital loss from multiple compression is.
Growth Quality1 · NoneRevenue $79.5M (2023) → $83.3M (2024) → $45.9M (2025), −44.9%; quarterly 2025: $14.0M → $14.5M → $10.1M → $7.3M, then Q1-26 $8.6M (−39% YoY). Consensus 2026 revenue ~$42.4M — below 2025 (FMP estimates, 6 analysts). FY25 gross margin 31.0% annually but swinging violently (Q3-25 −8.9%, Q4-25 +38.1%, Q1-26 −17.4%). There is no reported growth; there is reported contraction.
Exponential Potential5 · Moderate (real option, unpriceable odds)Consensus revenue path: $42.4M (2026E) → $72.9M (2027E) → $124.6M (2028E) → $200M (2029E) → $500M (2030E) — a hockey stick that, if real, is a ~64% CAGR off 2026E. The 800 VDC transition is a genuine architectural shift and Navitas has both GaN and SiC. But: one of 14+ named partners, zero quantified dollar content or share anywhere in the public record, demo boards not products, EBITDA still negative through 2029E on the Street's own numbers, and China undercutting the category. High ceiling, unknowable odds — a 5, scored honestly as optionality.

The three cases (our own scenario model — assumptions shown; each target is a ~12–18-month fair value). We deliberately do not attach probabilities: the base case is the expected path, so a weighted blend would just restate it with false precision. One warning before the table: with negative gross margin and negative FCF, no fair value here is DCF-defensible. These are scenario anchors — multiples on estimates plus the cash — for a stock whose price is set substantially by attention.

CaseKey assumptionsFair value
BullKyber ramps on the 2027 schedule, Navitas discloses a committed-volume design win with real dollar content, gross margin turns positive toward the ~40% management model, and the market pays for the 2029E-2030E path ($200M → $500M, estimates): EV ~$3.3B ≈ 16-17× 2029E sales.~$16 (+40%)
Base (our anchor)The 2028E consensus (~$124.6M revenue, still EBITDA-negative) roughly lands; the market pays a still-generous ~11-12× EV/2028E sales (~$1.4-1.5B EV) plus ~$230M net cash for the optionality.~$7.50 (−35%)
BearAnother negative-gross-margin year, Kyber slips to 2028 (as SemiAnalysis reports), estimates cut, attention rotates; the stock revisits its pre-story zone — the $5.44 52-week low is the attention floor, and even $5 still implies ~$880M EV ≈ 21× 2026E sales. Fundamentals alone would justify lower.~$5 (−56%)

Synthos fair value = the base case, ~$7.50 (−35%), with the full $5–$16 span as the honest range. When the base case sits 35% below the price and the bull case requires facts (dollar content, committed volumes, on-schedule Kyber) that do not yet exist anywhere in the public record, the verdict writes itself: Avoid. This is a tracked call — the Forecaster Scorecard grades it once it matures.

4. Exponential Potential

Synthos separates compounders (durable returns on real demand) from exponentials (accelerating, multi-baggers-from-here). Navitas is a pure option on an exponential, with no compounder underneath:

Exponential Potential: Moderate (5/10). A genuine call option on the 800V transition — but an option is what it is, and the market is charging ~60-80× sales for it after two failed breakouts.

5. Financials (real numbers — FMP annual/quarterly)

6. Valuation — priced in or room?

There is no way to make this cheap. On trailing sales (TTM revenue ~$40.5M), the stock trades ~69× P/S and ~64× EV/S (FMP TTM ratios; EV ~$2.58B after netting cash). Annualizing the latest quarter ($8.6M × 4 ≈ $34.4M) puts it at ~75× EV and ~81× market cap to run-rate sales. On forward consensus it barely improves: ~61× 2026E EV/S ($42.4M — again, below 2025), ~35× 2027E, ~21× 2028E — and those out-years are estimates for a ramp with no disclosed design-win economics. There are no earnings to multiply (TTM EPS −$0.60; consensus stays negative through 2029E). P/B is 6.1× against a book that is ~49% goodwill/intangibles. FMP's letter rating is C (overall 2/5; DCF, ROE, ROA and P/E sub-scores all 1/5 — only the debt score is good, because there is no debt).

Street context (treat with care): the consensus price target is $18.23 (high $21 / low $13.70, FMP), +59% above spot — but the ratings tell the truer story: 0 Strong Buy, 3 Buy, 4 Hold, 1 Sell — consensus "Hold." Targets set during the spring 2026 spike have not caught down to a stock that halved since; a sell-side that won't rate it Buy at −64% off the high is not actually bullish. For calibration, the market has already priced this story twice: +25% on the day of the May 2025 NVIDIA announcement, and ~+20-22% around Computex 2026 (closed +19.3% at $30.84 on June 3) with an immediate after-hours fade (Synthos verified-research run, 2026-07-19) — both spikes fully retraced. Bottom line: the story premium survives at ~60-80× sales even after the round trip; nothing in the reported numbers supports it.

7. Technicals (from the tech block)

8. Moat & competitive position

Navitas's claim to a moat is technology breadth in wide-bandgap power — one of few pure-plays with both GaN (GaNFast, monolithic integration) and SiC (GeneSiC) under one roof, and demo-level performance (97.5% claimed peak efficiency, 2,100 W/in³ on the 800V→6V board — company figures, not third-party validated). That is genuine engineering. It is not yet a business moat, for four reasons: (1) the partner list is the anti-moat — NVIDIA named 14+ silicon companies to the same 800 VDC program, including Infineon, TI, onsemi, ST, MPS, ROHM, ADI, Renesas, Innoscience, AOS, EPC, Power Integrations and Richtek; several have vastly greater scale, automotive-grade quality systems, and existing NVIDIA/hyperscaler relationships; (2) no quantified content — no source anywhere breaks out per-rack power-semi dollar content or Navitas's share of it; (3) trailing-edge economics — power semis are not leading-edge logic; Dylan Patel's KB-tracked claims (2026-02-03/05, thematic context — his text names TI/ST, not Navitas) describe China power chips "matching and undercutting" Western incumbents, with Innoscience the China GaN volume leader; (4) customer concentration of narrative — the equity story is one customer's architecture slide.

Peer set: FMP's peer list is noisy (it includes Grindr and Quantum Computing — ignore those). The relevant comps are Power Integrations (POWI, ~$3.9B cap) — profitable, at a fraction of NVTS's sales multiple — plus the diversified power incumbents (Infineon, onsemi, ST, TI) and private/China GaN (Innoscience, EPC). Against every one of them, Navitas is the smallest, least profitable, and most expensive per dollar of revenue.

9. Management, capital allocation & guidance

10. Catalysts & what to watch

Pre-registered flip conditions (what would move us off Avoid — these are the monitoring tripwires, logged 2026-07-19):

1. DC/800V revenue broken out in filings, or a design-win with committed volumes/dollar content disclosed;

2. Two consecutive quarters of positive, rising gross margin on a path toward the ~40% management model;

3. Kyber 2027 confirmed (production, not roadmap reiteration);

4. Forward consensus estimates inflecting upward (2026-27 revenue revised up, not just out-years);

5. No ATM/secondary dilution into strength across the intervening prints.

Any two of these moves us to Watch; the first plus the second together would trigger a full re-underwrite.

11. Key risks

(Risks to a holder — we are recommending against holding; these are why.)

12. Verdict, position sizing & monitoring

Avoid. The 800 VDC datacenter transition is real, and Navitas's GaN+SiC engineering is real. But an investment needs a business underneath the story, and the reported business is going the wrong way on every line that matters: revenue −44.9% in 2025 with consensus modeling 2026 lower still; gross profit negative in two of the last three quarters; ~$25-27M operating losses per quarter funded by issuing stock (+53% dilution since 2022); insiders selling nine figures at $28-32; and a valuation of ~60-80× annualized sales after a −64% drawdown. The base case fair value (~$7.50) sits 35% below the price, and the bull case requires facts — dollar content, committed volumes, on-schedule Kyber — that do not exist in the public record. When the market has already priced and un-priced this story twice inside a year, the burden of proof is on the numbers, and the numbers say no.


Provenance & disclosures