← Research/5/18/2026
Nefarious Trading
Skeptic's Take · discord.gg/nfrs
Vol. 01 · No. 14
May 02, 2026
⚠ SKEPTIC'S TAKE · THIS IS A TRADE, NOT AN INVESTMENT · READ THE FINE PRINT ⚠
KOPN $4.45 ↑ SBLX FABRIC.AI / EX-SILEXION CIEN $98.42 ↑ LITE $78.55 ↑ COHR $112.20 ↑ ANDU $45.30 ↑ KTOS $31.80 ↑ AVAV $215.40 ↑ KOPN $4.45 ↑ SBLX FABRIC.AI / EX-SILEXION CIEN $98.42 ↑ LITE $78.55 ↑ COHR $112.20 ↑ ANDU $45.30 ↑ KTOS $31.80 ↑ AVAV $215.40 ↑
$KOPN · Skeptic's Take · Defense + AI Hype Cycle

KOPN ran 70% in three weeks on an AI deal with a rebranded biotech. The fine print kills the moonshot.

$15M "purchase order" is actually $5M committed plus $5M in escrow tied to a prototype demo. Their "AI partner" is a former clinical-stage biotech that pivoted three months ago. Share count up 40% in one year. The COO sold one day after the catalyst. The trade is fine. The thesis is not.

Price
$4.45
From $2.58 on Apr 6
3-Week Run
+72%
Mid-cycle, not early
FY25 Revenue
$39.3M
Down 22% YoY
EV / Sales
~17x
Hardware peers at 4-5x
§ 01 — The Honest Setup

Three things the X bulls are not telling you. All three matter.

KOPN is being pitched as the next great AI infrastructure pure-play with a defense moat underneath. That framing has problems. The Fabric.AI deal is structured very differently than the headlines suggest. The AI partner itself is a rebranded micro-cap with no operating history in semiconductors. And the company has diluted shareholders aggressively at every previous run-up. The trade may work as momentum. The investment thesis does not survive contact with the 8-K.

The TL;DR: Real defense business shrinking 22% YoY. AI deal is mostly conditional milestone payments. Partner is a sub-$100M biotech that pivoted three months ago. Share count up 40% in 12 months. Insider sold 1 day after the catalyst. This is a momentum trade with a stop, not a buy-and-hold.
Red Flag 01
The "$15M deal" is mostly conditional
SEC 8-K shows $5M committed + $5M in escrow tied to prototype demo. Remainder is "expected" production-stage payments not yet contracted.
Red Flag 02
The AI partner is a biotech rebrand
Fabric.AI trades as $SBLX. Was Silexion Therapeutics, a clinical-stage cancer biotech, until pivoting and rebranding in early 2026.
Red Flag 03
40% share dilution in 12 months
Two equity raises in 2025 at $0.65 and $2.10. Share count went from 124M to 183M. 10-K explicitly warns more capital may be needed.
§ 02 — Decoding The Fabric.AI Deal

Read the 8-K. Not the press release.

The X thread reads "Kopin secures $15M order from Fabric.AI for AI optical interconnects" and your dopamine fires. The actual SEC filing is structured very differently. Here is what the Joint Development and License Agreement actually says, line by line:

ElementPress Release FramingActual 8-K Language
Initial commitment"$15M purchase order"$5M initial purchase order
Tranche 2(implied as same deal)$5M segregated funding account, contingent on at least one successful prototype demo
Tranche 3(rolled into the $15M)$5M further development funding, conditional
Production phase(not in headlines)"Expected to negotiate" $15-25M additional after demo. NOT signed. NOT contracted.
Equity stake"19.9% of Fabric.AI"Series J Convertible Preferred. 6% annual dividend payable in cash OR additional stock. Anti-dilution adjustments.
Manufacturing rights"Exclusive manufacturer"Exclusive only if production plan is successfully negotiated post-demo

The pattern is clear: the deal is gated by milestones, and the bigger numbers are conditional on a prototype that does not exist yet. If the demo fails or slips into 2027 (typical for first-generation chipset development), Kopin gets the initial $5M and not much else.

The honest framing: This is a $5-10M near-term development contract with optionality on a $15-25M production phase that has not been negotiated. The market is pricing it as if all $30-40M in potential payments are committed. They are not.
§ 03 — Who Is Fabric.AI, Actually

Their "AI semiconductor partner" was a biotech three months ago.

This is the part that should make you stop and think. Kopin's strategic AI partner is Fabric.AI, ticker $SBLX on Nasdaq. The X bulls describe them as "a leading developer of fabless semiconductor solutions for AI infrastructure." Pull up the ticker history and a different story emerges.

SBLX Corporate History

From Cancer Drugs to AI Chips in 90 Days
Before
Silexion Therapeutics
Clinical-stage biotech focused on RNAi cancer therapeutics. Trading as a tiny micro-cap with limited revenue and ongoing trial-stage drug development.
After
Fabric.AI
Rebranded as fabless semiconductor company for AI infrastructure. Sub-$100M market cap. No prior chip design history. No prior optical interconnect IP.

The pivot pattern: Small biotech runs out of clinical runway. Pivots to whatever narrative is hot (crypto in 2017, EV/SPAC in 2021, AI in 2024-2026). Rebrands. Lines up a partnership with a real operating company. Issues press releases. The stock runs. Insiders sell. The company that did the rebranding is rarely the one that builds the product.

What this means for Kopin: The 19.9% equity stake in Fabric.AI is being framed as ownership of a real AI semiconductor business. It is ownership of 19.9% of a tiny rebranded shell with one announced product concept and zero shipped revenue. If Fabric.AI cannot raise the capital to fund the production phase or fails to deliver a working demo, the partnership collapses and the equity stake is worth substantially less than the headline number.

What this does NOT mean: The technology itself may still be real. Kopin owns the underlying MicroLED IP and patents. The question is whether their commercial partner can execute. Today, that answer is unclear at best.

The skeptic's read: Real AI semiconductor partners look like Broadcom, Marvell, Coherent, Lumentum. Companies with operating history, real chip design teams, customer pipelines. Fabric.AI is none of these things. The partnership might still work out. But the X framing of "Kopin owns 19.9% of a real AI semiconductor company" is materially misleading.
§ 04 — The Defense Business

The "moat" is real but the revenue is shrinking.

Defense is 74% of KOPN revenue. This is the part of the business that has actual operating history and sticky relationships with the DoD and allied militaries. The bull thesis says this is a stable foundation that funds the AI optionality. The numbers say the defense business is contracting, not growing.

The Revenue Trend Tells The Story

MetricFY 2023FY 2024FY 2025Trend
Total Revenue$40.4M$50.3M$39.3M-22% YoY
Defense Product Revenuen/a~$37M$29.4M-21%
Defense % of Total~70%~74%74%Steady mix
Net Loss($19.7M)($43.9M)($6.2M TTM)Improving
Operating Cash Flownegative($14.2M)($15.5M)Worse
Accumulated Deficit~$355M~$393M$399.5MGrowing

Recent Contract Wins (Real, But Small)

The 2026 announced contracts are real but the dollar values are modest relative to the market cap:

ContractInitial ValuePotentialDate
Sentinel FPV drone goggle modules$3.2MUp to 40,000 units through 2028Apr 2026
Theon DarkWAVE 960p development order$1.0MProduction-readiness in 2026Q1 2026
Tier-1 EU defense contractor (helmet HMD)$2.0MMulti-year follow-on potentialFeb 2026
Pilot helmet-mounted display$2.0MMulti-year programFeb 2026
EU contract (multiple systems)$3.6Mn/aFeb 2026
SBIR Phase I (MicroLED soldier displays)under $1MPhase II contingentApr 2026

Total announced 2026 contract wins: roughly $12M committed + production options. This is not nothing. But for a company with an $815M market cap, these wins do not justify the rerate on their own. The market is pricing in much larger production-phase ramps that have not yet been signed.

The honest read on defense: Real moat, real customers, real contracts, but a contracting top line and rising operating losses. The 2026 wins are encouraging but small. Production-phase economics on these programs will not show in revenue until 2027 at the earliest.
§ 05 — The Dilution Pattern

Every prior run-up funded a raise. Why would this one be different?

This is the math the bulls do not want to do. KOPN has financed itself through equity issuance for years. Every previous catalyst has been followed by a raise at progressively higher prices until the price comes back down. The pattern is so consistent it is the base case.

FY 2023
~110M shares
FY 2024
~124M shares
FY 2025
173M shares
Apr 2026
183M shares
Risk: Mid-2026
200M+ likely

The Two 2025 Raises

Raise 1: $33.9M of stock and pre-funded warrants at $0.65 per share

Raise 2: $38.1M private placement at $2.10 per share

Combined: $72M in new equity issued in one year, increasing share count by 39.5%.

At $4.45, every existing investor who participated in those raises is sitting on 110-580% gains. The math problem: the company burned $15.5M in operating cash flow in 2025, has $61.6M cash on hand, and has explicitly warned in its 10-K that additional financing may be required if profitability is not achieved. Translation: another raise is highly likely, and the stock at $4+ is the perfect price to do it.

The dilution math nobody wants to do: If KOPN raises another $50M at $4.00 (a typical small-cap discount on overhead capital), that adds 12.5M shares (+7%). The share count goes from 183M to 195M. The market cap impact at $4.45 = $55M dilution against existing holders. This is the most likely scenario in the next 6 months. Plan for it.

The Insider Sale

On April 29, 2026 - one day after the Fabric.AI announcement pumped the stock - COO Paul Christopher Baker filed a Form 4 disclosing the sale of 116,860 shares for approximately $462,000 at $3.98. He still holds 480,005 shares.

Insiders rarely time perfectly. They almost never sell into news they think is multi-bagger material. Selling 20% of your position one day after the company's biggest catalyst announcement is the strongest possible signal that the COO believes the stock is closer to fair value than the bulls think it is.

§ 06 — Scorecard

Trade thesis vs. investment thesis. Two very different conclusions.

Bull / Trade Case

  • Real momentum. 70%+ in 3 weeks with volume confirmation. Trend is intact above $4.
  • Multi-analyst initiations. Stifel $5.50 Buy. JonesResearch $6 Buy. Lake Street $5 raised from $4.
  • Two real narrative tailwinds. Defense modernization + AI infrastructure. Both have strong sponsorship.
  • Cash cushion. $61.6M in cash gives them 18-24 months of runway at current burn.
  • Real defense moat. Sole-source on multiple DoD programs. Sticky IDIQ contracts. ITAR-free DarkWAVE export potential.
  • Optionality on Fabric.AI. If the demo succeeds, the production phase could materially change the revenue trajectory.
  • Short interest at 9.25%. Squeeze potential on any positive surprise.

Bear / Investment Case

  • Revenue declining 22% YoY. The "ramp" is forward-looking. The reported numbers are going the wrong way.
  • Fabric.AI deal is mostly conditional. $5M committed, $5M escrow, rest is "expected to negotiate." Read the 8-K.
  • AI partner is a 90-day biotech rebrand. $SBLX = ex-Silexion Therapeutics. Sub-$100M market cap. Zero chip design history.
  • 40% share count increase in 12 months. Pattern of dilution at every previous run-up. Another raise is highly likely.
  • COO sold 20% of position 1 day after the catalyst. Strongest possible insider signal.
  • 17x EV/Sales for shrinking hardware revenue. Optical peers (Ciena, Lumentum) trade at 4-5x. The premium is pure narrative.
  • $399.5M accumulated deficit. Decades of losses. 10-K explicitly warns more capital may be needed.
  • Multiple "contract wins" total under $12M. Real but small. Production economics are 2027+ stories.
The honest framing: The bull case is a momentum trade with technical support. The bear case is a fundamental investment thesis. They are not mutually exclusive in the short term but they fully diverge over 6-12 months. Treat this as a trade. Do not marry it.
§ 07 — Price Scenarios

Where it can go. And where it probably stops.

Analyst targets cluster at $5-6. The X bulls are calling for $10-15-20. The reality is most of the easy multiple expansion has already happened. Here is the scenario map:

Bear · 3-6mo
$2.50
-44%
ATM offering announced. Defense Q1 2026 misses. Fabric.AI demo slips into 2027. Multiple compresses to 5-6x sales. Most likely outcome if any of three risks hit.
Base · 3-6mo
$5.00
+12%
Stifel/Lake Street target zone. Stock holds the breakout, defense stabilizes, demo is on track. Modest upside from current price. Most likely if execution holds.
Bull · 6-12mo
$7.00
+57%
JonesResearch target. Demo succeeds. Fabric.AI production phase signed. Defense Q2-Q3 reaccelerates. Requires three things to go right.
Stretched · 12-18mo
$10+
+125%
X moonshot zone. Requires Fabric.AI to be a real semiconductor company AND no further dilution AND defense ramps to $80M+ revenue. Possible but priced as base case is wrong.

Why $20-30 Targets Don't Math

For KOPN to hit $20, market cap needs to be $3.7B+ on current share count, or $4B+ assuming continued dilution. To justify that on revenue, FY27 needs to be $400-500M (10x current). Even the most aggressive bull model projects $80-120M by 2027. The math gap is 4-6x. The X "$20-30 by year end" calls require either a buyout (no current rumors) or a momentum overshoot that historically corrects within weeks.

Realistic ceiling: $6-7 over 6-12 months if execution holds. Anything above $5 is gravy. Anything above $7 should be sold aggressively. The asymmetry is no longer favorable above $5.
§ 08 — My Take

The trade is fine. The thesis is broken.

KOPN is the kind of small-cap where the headline narrative and the SEC filings tell two different stories. The narrative says: defense moat plus AI optical interconnects equals multi-bagger. The filings say: shrinking revenue, conditional milestone payments from a rebranded biotech, 40% dilution per year, and a COO selling the day after the catalyst.

This does not mean the stock is a short. Momentum is real. Analyst initiations are flowing. Short interest at 9.25% creates squeeze potential. The defense contracts are legitimate even if small. And if the Fabric.AI demo actually works, there is a real (low-probability, high-impact) path to a much higher price in 12-18 months.

What it means is this is a trade with a hard stop, not a position you sleep on. The bulls on X are conflating "momentum is working" with "thesis is correct." Those are different statements. Momentum can extend further than fundamentals justify. But when momentum breaks, the fundamentals reassert themselves quickly, and the fundamentals here have a 22% revenue decline, a sketchy AI partner, and a near-certain dilution overhang.

Trading KOPN long with a hard stop at $3.50.

Entry zone: NOT chasing here. Wait for either (a) pullback to $3.80-4.00 with volume contraction, or (b) clean break above $5.00 with continuation. Trim 50% at $5.50 (Stifel target). Trim another 25% at $6.00. Cut 100% on any ATM offering announcement, any failed demo timeline update, or any break of $3.50. Adding only on confirmation of Fabric.AI prototype success - not on speculation. Position size SMALL. This is a trade, not a thesis position.

"This is the type of stock where reading the 8-K saves you 40%. The X thread that didn't is the one you remember."