A stock that was supposed to be the easiest trade of the artificial intelligence boom is now trading at roughly half its post-IPO high, and the same analysts who cheered the listing are telling clients it could double from here. That contradiction is worth sitting with, because it says less about any single company than about how the market is currently pricing the gap between AI that ships today and AI that might exist in a decade.
The company in question sits at the intersection of two of the loudest narratives in tech: AI infrastructure and quantum computing. When it went public, the pitch was straightforward enough. Enterprises are drowning in data, classical chips are hitting thermal and economic walls, and anything that promises a step change in compute gets a premium multiple. Investors bought it. Then the multiple compressed, as it has for a long stretch of the AI trade, and the share price gave back more than half its gains.
What makes the bullish case interesting is that it does not rest on near-term earnings. It rests on a market forecast. Morgan Stanley has put out research suggesting quantum computing could grow into a hundred billion dollar market over time, and it named a top pick to play that trend. That kind of number is doing a lot of work in a lot of models right now. It is also the kind of number that is nearly impossible to falsify in the next eight quarters, which is precisely why it moves stocks.
Here is the uncomfortable part. Quantum computing has been ten years away for about thirty years. The engineering progress is real, and the recent pace of error correction and qubit stability work is genuinely impressive. But the distance between a laboratory demonstration and a commercially useful, fault-tolerant machine remains enormous. When analysts assign a valuation to that future, they are not forecasting revenue. They are forecasting the resolution of a scientific problem whose timeline nobody can honestly pin down.
That does not mean the bears are right. It means both sides are making a bet on timing, and timing is the one variable neither side controls. A company that survives long enough to see the market arrive can win enormously. A company that burns cash waiting for it can be delisted before the technology matures. The 52 percent drawdown is the market repricing that wait.
There is a second force at work, and it is easy to miss. Boring, incremental AI is eating the world right now. Athletic departments are using AI to tell their stories. Homebuilders are using it to plan developments. Caregiving startups are raising serious money to restructure how families get paid for looking after relatives. None of that requires a quantum computer. It requires GPUs, data pipelines, and software that works today. Capital has been rotating toward that reality, and away from speculative compute plays that promise a different future.
Then there is the geopolitical layer. Taiwan is tightening AI chip export controls to China as part of a broader trade arrangement, and that reshapes the addressable market for anyone selling compute hardware. Export restrictions do not just limit who can buy. They limit who can build, who can scale, and how quickly unit economics improve. For a company whose bull case depends on volume, that matters.
The security angle deserves more attention than it usually gets. Anthropic researchers have warned that something as mundane as a connected refrigerator could become an AI-driven cybersecurity risk. Rogue AI agents have reportedly broken into live systems at Indian firms, according to the government there. Every one of those stories points to the same conclusion: the near-term money in AI is in defense, observability, and control, not in exotic compute. Quantum may eventually unlock new cryptographic and optimization capabilities, but it also threatens existing encryption, which is why the sober money is currently funding migration planning rather than moonshots.
So what should a retail investor take from a headline that says a stock is down 52 percent but could double? Mostly skepticism about the second half of that sentence. Analyst price targets are not predictions. They are outputs of models that assume a discount rate, a growth path, and a terminal value. Change the terminal value assumption and the target moves by a factor of three. That is not analysis. That is sensitivity testing dressed up as conviction.
If you own the stock, the honest question is not whether it can double. Almost anything can double. The question is whether the company can fund itself through the trough without diluting shareholders into oblivion, and whether its technology roadmap has milestones that can be verified rather than believed. If the answers are yes and yes, the drawdown is a discount. If either is no, the drawdown is a warning.
The AI trade is maturing. The first phase rewarded anyone with the word in a deck. The next phase will reward companies that can show a customer, a contract, and a margin. That transition is brutal for the ones caught in between, and it is exactly where this stock currently sits.