$QUIK is up ~274% in a year on a real defense story. The tech is legit. The price is not.
No position in $QUIK. Won't trade 72hrs after posting. Analysis not advice. DYOR. This post is not investment advice.
I built the model before chasing the chart, and the math says the market is paying for a near-perfect outcome the numbers do not support. Here is the financial case.
WHAT THE COMPANY IS
A fabless semiconductor shop that licenses embedded FPGA blocks as silicon-proven Hard IP, and is now launching its own radiation-hardened FPGA products for defense and space. It holds one genuinely rare asset: the first and only eFPGA Hard IP on Intel 18A, the leading-edge US foundry node.
THE Q1 2026 PRINT (reported May 12)
Revenue $5.05M, up 16.8% YoY and 35% sequentially
New products 85% of sales
GAAP gross margin 36.5%
GAAP net loss $2.2M, or -$0.13
Non-GAAP loss $0.08, missed the ~$0.05 estimate
The revenue beat was small, the loss ran wide, and margin compressed year over year as mix shifted toward early product.
THE NUMBER NOBODY IS WATCHING
Cash fell to $6.0M, from $18.8M at year end. There is a $125M shelf and a $20M at-the-market program sitting over the stock. For a company burning roughly $2M a quarter, dilution is not a tail risk. It is the likely funding path for the next phase.
WHY NOT A NORMAL DCF
The core eFPGA and licensing business does not earn enough to cover its own ~$16M operating expense base. A standard five-year DCF just burns cash and buries every heroic assumption in terminal value. So I valued it the honest way, as a core business plus three separate call options, risk-adjusted. An rNPV.
THE rNPV (all discounted at 14%)
Core business: ~$60M, or about $3.30 per share
Storefront rad-hard products, 35% odds: ~$42M
Intel 18A first-mover position, 30% odds: ~$45M
SRH government expansion, 45% odds: ~$27M
Base case ~$9.50. Bull ~$17.50. Bear ~$4.20. The stock is at ~$22.50, above even the bull case.
A Graph of the rNPV is attached to this post.
THE MONTE CARLO (100,000 runs)
I stress-tested it, letting each option hit or miss and each payoff vary, with dilution rising as options miss.
Median outcome: $8.26
Below $5: 27% of the time
Above the $22.50 market price: just 1.9%
That last number is the whole thesis. Across 100,000 runs of a model that gives QuickLogic full credit for three real options, only 1.9% of outcomes justify today's price. The market is not buying the middle of this distribution. It is buying the far right tail.
A graph of the Monte Carlo is attached to this post.
THE HONEST READ
The grounded analysts agree with the base case. Oppenheimer sits at $11. The low-$20s targets only moved up after the stock did, not before.
Real tech, a defensible niche, and a rare Intel 18A edge I respect. But the core does not self-fund, the entire equity value is a stack of options, and dilution is the plan rather than the risk. Fair value lands near $9.50. I would own this at $8 to $12. At $22.50 you are not buying a business, you are buying a clean sweep.
No position. Watching for a better entry.
Full breakdown is on my Substack: what an eFPGA actually is, how radiation flips a bit and why rad-hardening is hard, the Intel 18A angle, the complete rNPV and Monte Carlo, and every risk. Link on my profile.
Not investment advice, only for entertainment.
@BamaBonds I feel less bad for retail idiots who choose to buy it, but the potential of having it crammed into indexes early was disturbing. You'd have a wide swath of people who didn't even realize what they were buying at what price. Glad S&P at least didn't change rules.
@DanielleFong This points to a resilient labor market and supports higher-for-longer rates. Result: elevated discount/hurdle rates for long-duration tech bets. Makes the risk-adjusted return math harder for tech. Tech goes down, paper handed tech holders sell, positive feedback back loop.
After watching what happened to under armor, I decided to never again be in apparel space. You summed up the issues very well. There is no switching cost or friction to just buy vuori over lulu and in a couple years no one will know who vuori is and they'll probably become vuori AI and rent gpus
Smaller open source models that are tuned for specific tasks seems like the way forward. Ran a home llm set up with qwen-coder and the results were very good compared to my expectation. I think what we'll see in the future is that models going after AGI price themselves out of the market while models that do one thing really well will wind up seeing adoption.
May payrolls came in at +172k (vs ~85k expected), with sizable upward revisions to prior months. This points to a resilient labor market and supports higher-for-longer rates. Result: elevated discount/hurdle rates for long-duration tech bets. Makes the risk-adjusted return math harder for tech. $NVDA $AVGO $MU
Its interesting too, because they actually held their full year AI numbers. They didn't cut their guide, the just didn't raise it. The overall numbers would have looked insane a year ago. Investors pricing in continuous acceleration of spending, which is unrealistic. Or just people who had one foot out the door to begin and took this time to profit take.
Yeah that is the thing, a lot of people have started using AI for every little thing and question as opposed to where it actually adds value. Its still a very young technology and companies and employees will need to adjust their expectations and usages to what it can actually do well and cost effectively. Tokenmaxxing (really any of these maxing trends) was always a silly idea.
@unusual_whales Honestly i think Trump will try to take this out. He need a way out of the war while saving face in order to keep gas / inflation low and stock market high. If he has a chance to end the war and blame democrats in congress I am willing to bet he takes it
@KobeissiLetter Honestly i think Trump will try to take this out. He need a way out of the war while saving face in order to keep gas / inflation low and stock market high. If he has a chance to end the war and blame democrats in congress I am willing to bet he takes it
Yeah I tend to agree with this. I see a lot of stuff about how expensive AI is per token and I agree the cost will absolutely need to come down, but for these 500 million type stories, im gonna need a receipt and be told what the heck they were doing to even use that many tokens. I use AI a lot at the day job and I just can't see that happening.
Its interesting though, and I'm not saying AI won't displace jobs, but today's jobs report showed a record amount of openings it hasn't started to displace works yet. I wonder if it is better to wait to do something this drastic until you start seeing mass displacement or firing of workers. It would be a very big departure from the way American has let companies run before.
@Alisvolatprop12 If the tool truly can find vulnerabilities that have been present for decades that no traditional security team was able to find, then it seems that regardless of cost major software companies will have to use it or risk having leaks and outages that ruin their reputation