The best-placed independent maker of display glasses, priced for a future my model cannot reach on its own. I take the bet as a conviction call and show the numbers that argue against it.
How to read this report. Evidence tags: (A) reported by press or a registry, (C) stated by the company and unaudited, (E) third-party estimate such as Counterpoint or IDC, (D) my own derivation. The reference price is the $1.0B post-money value of the July 2026 funding round. It is private and illiquid, so there is no live price.
Prices from heise, The Shortcut and TechCrunch. Product and supply chain
This is a conviction call, and the base case does not support it. The headline Bullish rating rests on the conviction case below, worth about $1.33 billion of equity against the $1.0 billion July 2026 round. My probability-weighted model is worth $0.40 billion, and my base case alone $0.28 billion, both well below the round. For a reader this means a high-risk, lopsided bet: the payoff needs Even to keep its share of display glasses as the largest technology companies arrive, and if it cannot, most of the value paid in July is not supported by cash flow.
Meituan, Tencent and existing investors paid about $150 million for roughly 15% of the company at a $1.0 billion post-money value (A). My model value is 60% lower. The gap is not a different view of the product; it comes from three choices.
I discount at 20.2%: a 14.2% market cost of equity for a beta of 1.8, plus 6 points for being private, three years old and dependent on one product family. Strategic investors buying an option on a future platform can accept far less. At 14.2% my model value would be $0.53 billion, still about half the round.
Meta sells Ray-Ban Display for $799 and now a camera-free Ray-Ban Meta Audio for $349. Google and Samsung have shown Android XR glasses, and Apple’s are expected in late 2027. My base case lets Even’s share of display glasses fall from 11% to under 4% by 2035, while its units still grow eightfold.
A three-year-old company will refresh its option pool and probably raise again before any listing. I take 12% off the base case value for today’s holders (10% bull, 25% bear). Investors who price a round on post-money value alone skip this step.
Even Realities is Shenzhen Yiwen Technology Co., Ltd. (深圳市逸文科技有限公司), incorporated in Shenzhen on June 2, 2023 (A). It designs display glasses that look like ordinary frames: a green monochrome micro-LED projector in each arm, a waveguide in each lens and a phone app that drives translation, navigation, a teleprompter, notifications and a conversation assistant. The G1 launched in June 2024 at $599; the G2 and the R1 ring followed in November 2025. It builds in its own plant in Guangzhou, sells in more than 40 countries through its website, more than 300 European opticians and retailers in Japan, Korea, Taiwan and Hong Kong, and does not yet sell in mainland China (C).
It publishes no financial statements. The table shows my estimates next to the few numbers the company or the press have disclosed.
Estimated history (D), USD millions unless noted
| Line item | 2024E | 2025E | 2026E | Disclosed anchor |
|---|---|---|---|---|
| Pairs shipped (thousands) | 8 | 18 | 65 | ~3,000 G2 by end-Jan 2026 (A/C); “more than 10,000 pairs” by Jul 2026 (C); 11% of display glasses in H1 2026 (E) |
| Net revenue per pair ($) | 640 | 700 | 780 | Average order ~$1,000 (C); RMB 16M on ~3,000 pairs implies ~$750 (D) |
| Revenue | 5 | 13 | 51 | Not disclosed |
| Gross margin | 40% | 42% | 46% | Not disclosed; XREAL 35.2%, Garmin 58.7% (A) |
| Operating expenses | 12 | 18 | 26 | Headcount 30–40 (2024) to 300–400 (2026) (C) |
| EBIT | (10) | (13) | (3) | “The company is profitable” (C, unaudited, basis not stated) |
Sources: Jiemian via Sina Finance (Apr 17, 2026) for G2 deliveries and order revenue; TechCrunch (Jul 6, 2026) for the 10,000-pair claim, profitability, average order and headcount; Counterpoint (Sep 21, 2026) for share. Estimates are mine; the 2024 and 2025 figures are the least certain numbers in this report and do not drive the valuation.
Even’s design choice is to leave things out. No camera means no recording light, less weight and longer battery life; no speaker means the glasses are for reading, not listening. The company calls it “camera-free by design” and markets it directly against Meta’s camera glasses.
| Product | Launch | Price | What it is |
|---|---|---|---|
| Even G1 | Jun 2024 | $599 | First model: binocular JBD micro-LED, 640×200, 25° field of view, about 1.5 days of battery (A) |
| Even G2 | Nov 2025 | $599 · €699 | 36 grams, 640×350, 27.5°, 1,200 nits, IP67, about 2 days of battery; no camera, no speaker (A) |
| Even R1 ring | Nov 2025 | $249 · €269 | Controls the glasses by touch and tracks health; about 4 days of battery (A) |
| Prescription lenses | with G1/G2 | $159–200 | −12 to +12 dioptres with astigmatism; single vision only (A) |
| Software | ongoing | free | Conversate, Prep Notes, translation, teleprompter, navigation, Even AI; Even Hub plug-ins since Mar 2026. No paid tier found (A/C) |
Sources: heise (Nov 13, 2025); The Shortcut G2 review (Jul 1, 2026); VRAR星球 via Tencent News (Sep 29, 2024); Jiemian via Sina Finance (Apr 17, 2026); China Fund News (Jul 6, 2026); Even Realities company page and specs.berlin partner post (Dec 31, 2024); Wellsenn XR Rokid Glasses teardown (Feb 24, 2026); Engadget, Meta Connect 2026 (Sep 23, 2026).
Even’s 11% share in H1 2026 came while Meta’s Ray-Ban Display was supply-limited (its waveguide orders reached about 150,000 units, TrendForce) and before Google, Samsung and Apple ship display models. Each point of share in 2035 is worth about 150,000 pairs a year in my market size.
The G2 holds $599, and lenses and the ring lift an average order to about $1,000 (C). IDC expects prices of display-free glasses to fall about 40% by 2030. My base case lets Even’s net revenue per pair fall from $780 to $620.
Margin is the profitability claim. I assume 45% to 46% in the base case and 50% in the bull, against XREAL’s 35.2% (2025) and Garmin’s 58.7%. JBD’s move to 12-inch wafers and a local waveguide supplier should lower costs; price competition pushes the other way.
Even does not sell in mainland China, where 45% of display glasses shipped in H1 2026 (Counterpoint) and the government now subsidizes smart glasses. Meituan and Tencent can open that market. A large launch would need spending, and probably more equity.
The model is driver-based: pairs shipped times net revenue per pair, a small software line, operating expenses in dollars including stock compensation, cash tax at China’s 15% high-tech rate on positive profit, then unlevered free cash flow through 2035. Every driver is my estimate (D).
Base case, USD millions (2026E is my estimate)
Scenario values are equity value to today’s holders in USD billions, after each scenario’s expected dilution. Weights: bear 25%, base 50%, bull 25%.
The CEO says Even is profitable (C). With no accounts, the best test is the economics of one pair. All figures below are my estimates (D) unless marked.
Revenue Costs Operating costs, mostly people and marketing Result
My reading: plausible at the gross level, unproven for the company. The product earns a healthy margin, and operating costs per pair fall quickly with volume. That makes Even different from XREAL, which lost RMB 456 million in 2025 on RMB 516 million of revenue (A). It also means the valuation depends almost entirely on volume, which is why units are the driver I solve for in the reverse DCF.
Sources: TechCrunch (Jul 6, 2026) for average order and the profitability claim; Wellsenn XR (Feb 24, 2026) for the Rokid Glasses bill of materials; XREAL Hong Kong listing application (Apr 1, 2026) via 36Kr for its 2025 results.
Even has no public accounts, no trading price and a short history, so multiples of its earnings are not available. Its value depends on how many pairs it sells over the next decade and at what margin, which calls for a cash-flow model built from those drivers, run under several scenarios and checked against what similar companies sell for.
Pairs × net revenue per pair, software revenue, gross margins and operating costs, year by year.
Why: units and margin are what new share data and any disclosure will update.
EBIT less cash tax, plus depreciation, less capital spending and working capital. Stock compensation stays a cost.
Why: nine years reaches a mature margin in every case.
CAPM 14.2% (5.2% risk-free, beta 1.8, 5.0% premium) plus a 6.0-point private-company premium: 20.2%. No debt, so this is also the WACC.
Why: a private, three-year-old, single-product company carries risks a beta does not capture.
12x 2035 EBITDA, valued at the end of 2035, with a 3% perpetuity as a cross-check.
Why: a mature premium-hardware brand trades between Sonos and Garmin.
Bear 25%, base 50%, bull 25%. Each value is cut by its expected future dilution.
Why: a post-money value overstates what today’s holders keep.
Bullish if the 5-year expected return beats the cost of equity by 2 points a year. The conviction case is shown beside it, not mixed in.
Why: the reader can see where judgment overrides the rule.
| Method | How I used it | Reason |
|---|---|---|
| Driver-based DCF, exit multiple, three scenarios | Primary | Ties value to units and margin, the variables that new data will update. |
| DCF with perpetual growth | Cross-check | Gives $0.21B for the base case after dilution: a steady-state formula applied to a company still growing fast in 2035. |
| EV / sales against XREAL, Garmin, Sonos, GoPro | Cross-check | The multiples range from 0.35x to 13.8x, so the answer depends on the peer. Only XREAL’s private round supports $1.0B. |
| Reverse DCF | Cross-check | Solves for the unit path the July round implies: 2.3 times my base case. |
| Venture method (exit value discounted at a target return) | Not used separately | It is the same arithmetic as the exit-multiple DCF with a high discount rate, but hides the cash flows. |
| P/E or EV / EBITDA today | Not usable | No reported earnings; my 2026 EBIT estimate is about zero. |
I discount unlevered free cash flow for the fourth quarter of 2026 and for 2027 to 2035 at 20.2%, with mid-year discounting, and capitalize 2035 EBITDA at 12x as a value at the end of 2035. Net cash is about $0.17 billion (D): the $150 million round plus about $20 million of earlier cash. Each scenario’s value is then reduced by its expected dilution. The probability-weighted result is $0.40 billion against the $1.0 billion round.
| Peer | Sales multiple | Gross margin | Implied Even value |
|---|
Implied value: multiple × base-case 2027E revenue ($84M, D) plus net cash, less 12% dilution. Market caps from Robinhood market data at the Oct 16, 2026 close (price-to-sales; Garmin and Sonos hold net cash, GoPro has convertible debt). Bases: Garmin and Sonos FY2025 revenue; GoPro four quarters to June 2026; XREAL: about $1B round (Bloomberg, Jan 8, 2026) on 2025 revenue of RMB 516M from its Hong Kong listing application, at RMB 7.1 per dollar.
Holding every other base assumption fixed, the base case is worth the $1.0B round only if Even ships 2.3 times my unit path.
Comparable private and public marks: XREAL about $1B (Jan 2026, A) and still loss-making; Rokid reported at about $1B from its 2023–2024 rounds and preparing a Hong Kong listing (A); Meta’s Reality Labs lost $4.62 billion in Q2 2026 on $431 million of revenue (A); Meta paid about €3 billion for just under 3% of EssilorLuxottica in 2025 (A).
The conviction case runs the same model as everything above. It starts from my bull case and changes two inputs: 20% more pairs for a China launch, and a 14x exit multiple instead of 12x. The 20.2% discount rate and the 10% dilution stay. The result is $1.33 billion of equity value for today’s holders, about 9 times 2027E revenue at the enterprise level, below XREAL’s private multiple. Under the same rule it returns about 27% a year against a 20.2% hurdle, so it reads Bullish on its own terms.
| Belief | Conviction case | My base case | Evidence today |
|---|
Values from the model (public/even-realities-model.mjs, CONVICTION) and the Excel Conviction sheet, which recomputes the case with live formulas. The range on the football field is the conviction case at a 12x to 16x exit.
Each slider applies to all three scenarios. The scenario values, the model value and the rule update instantly; the published call does not change.
Even is run by its founder, with eyewear designers from two of Europe’s best-known frame brands. Its owners are mainly Chinese venture funds, with two Chinese internet companies added in July 2026. No board composition or ownership percentages are disclosed.
Founder and Chief Executive Officer; legal representative
What it means: a hardware operator, not a research founder. The Apple manufacturing background fits a company that built its own plant within three years.
Principal eyewear designer
What it means: the frames are designed by eyewear people, which is the core of the “glasses first” position.
Former LINDBERG chief marketing officer
What it means: the optician channel (300+ European stores) reflects relationships from the premium frame trade.
| Round | Date | Amount | Investors |
|---|---|---|---|
| Angel | Jan 2024 | n/d | Not disclosed (A) |
| Pre-A | Aug 2024 | n/d | Boyuan, Huachuang Capital, ChinaVenture fund (A, trade press); registry also added an Aier Eye affiliated fund |
| Series A | Dec 2024 | ~$9M | CDH Investments, Cyanhill Capital, China Growth Capital (A); Monolith named in one source |
| Unnamed | Jan 2026 | n/d | Unicorn Capital Partners (Hong Kong), Cyanhill Capital (A, secondary source) |
| Pre-Series B | Jul 6, 2026 | $150M | Led by Meituan (Longzhu and Strategic Investment); existing investors including Tencent and HSG (formerly Sequoia China) followed on. About $1.0B post-money (A) |
Sources: TechCrunch and China Fund News (Jul 6, 2026); Road to VR (Jul 6, 2026) for the $9M Series A and $159M disclosed total; VRAR星球 (Aug 13 and Sep 29, 2024) for registry changes; OFweek (Jul 7, 2026) for the January 2026 round. Tencent is described as an earlier backer; the date of its first investment is not public.
| When | Event | What I am watching |
|---|---|---|
| Feb–Mar 2027 | Counterpoint, IDC and Omdia full-year 2026 data | Even’s display share for H2 2026, after Ray-Ban Display’s wider rollout. Above 10% supports the conviction case; below 8% is a warning. |
| 2027 | Mainland China launch | Timing, price and whether Meituan or Tencent channels are used. The conviction case needs it. |
| 2027 | Google and Samsung display glasses | Price and iPhone support. A display model near $600 is direct competition. |
| After 2026 | Next-generation G3 platform | The company says a G3 is in development (tier-3 source). A new display or optics design would test the margin assumption. |
| Late 2027 | Apple glasses (reported) | Bloomberg reports no display in the first version. A display later would raise the bar for everyone. |
| 2026–2027 | XREAL and Rokid listings in Hong Kong | The first public prices for display-glasses makers; they would reset private marks for Even. |
Even reports nothing, so the next check comes from the trackers early in 2027. Here are my expectations, so the call can be tested against them.
Download the Excel model. It recomputes the three scenarios, the conviction case, the rule, a 30-cell sensitivity grid and the EV/sales cross-check with live formulas. Every hardcoded input carries a source note with its evidence tag, and it reproduces this page’s numbers exactly.
This is personal research by Phat Ngo, prepared as a work sample. It is not investment advice or a recommendation to buy or sell any security. Even Realities is a private company; its shares are not publicly traded, and the author has no position in it. The author’s portfolio shown on this site holds NVDA (10% target weight), AMD (9%) and SPCX (10%), among other positions. The headline Bullish rating is the author’s stated conviction and rests on the conviction case; the disciplined model reads Bearish at the July 2026 round value, and both are shown. Revenue, units and margins for Even Realities are the author’s estimates; the company does not publish them. Forecasts, scenario probabilities and the rating are judgments and may be wrong.