It's a highly risky bet, but not fundamentally unreasonable. One might believe that Ilya's research was genuinely critical to OpenAI's current situation. If one takes that premise, three potential corollaries follow: (1) OpenAI will struggle to produce future research breakthroughs without Ilya; (2) OpenAI will struggle to materially move beyond its current product lineup and variations thereof without said future research breakthroughs; (3) a startup led by Ilya could overcome both (1) and (2) with time.
An alternative sequence of reasoning places less emphasis on Ilya specifically and uses Ilya as an indicator of research health. Repeat (1), (2), and (3) above, but replace "Ilya" with something like "strong and healthy fundamental research group". In this version, Ilya's departure is taken as indication that OpenAI no longer has a strong and healthy fundamental research group but that the company is "compromised" by relentless feature roadmaps for current products and their variations. That does not mean OpenAI will fail, but in this perspective it might mean that OpenAI is not well positioned to capture future research breakthroughs and the products that they will generate.
From my perspective, it's just about impossible to know how true these premises really are. And that's what makes it a bet or gamble rather than anything with any degree of assurance. To me, just as likely is the scenario where it's revealed that Ilya is highly ineffective as a generalist leader and that research without healthy tension from the business goes nowhere.