OptQC's $47M from NTT buys a 10,000-qubit attempt, not a customer

The specification gap is the part worth reading closely. OptQC's published roadmap describes MoQuren as a 100-input quantum-analog demonstration machine operating at 100 MHz ^[11]. The successor is described as approximately 10,000 qubits at 100 times MoQuren's computational performance ^[3]. Multiply 100 inputs by 100 and you get 10,000 ^[23], which reads as a scaling of the same design rather than a new one. Inputs and qubits are not the same unit, and the mapping between them has not been published ^[4].

The scaling story rests on the architecture. OptQC generates optical pulses sequentially, sends them into the processor at different times and measures them in sequence, using quantum teleportation and time-domain multiplexing to perform operations with fewer optical components ^[13]. That is how a 100x step becomes arguable on paper: you buy scale in the time domain instead of in component count. It also means the timing and stability chain has to hold across a hundredfold longer sequence, and stability is precisely what OptQC itself lists as the first order of work on MoQuren, alongside classical integration and a usable developer environment ^[15].

The research lineage is deep. Furusawa demonstrated deterministic quantum teleportation at Caltech in 1998 and later built a teleportation-based method for optical quantum computers, according to OptQC's company profile ^[17], and Asavanant demonstrated a two-dimensional optical cluster state in 2019 ^[18]. That history answers whether the physics works. It does not answer whether a box stays aligned in a shared facility, which is the question the room-temperature, atmospheric-pressure design puts on the table when it drops the dilution refrigerator ^[14]. OptQC's argument that this cuts energy use and control complexity and eases data-centre integration ^[25] is an operations claim, and operations claims are settled by maintenance records nobody has published.

Now the money. Before this round OptQC had 2.15 billion yen of equity, roughly $14 million, from a 650 million yen seed and a 1.5 billion yen Series A1 ^[21]. The A2 is about 4.7 times the A1 ^[20] and about 76 percent of every equity yen the company has raised ^[19]. It also arrives roughly two years after incorporation ^[24], for a company formed in September 2024 in Toshima-ku ^[5] by a founder with eight years in the Furusawa Lab behind him ^[6]. Separately, the $133 million cumulative figure OptQC cites includes public research grants and should not be read as equity ^[9], which leaves about $72 million outside the cap table ^[22].

What exists to be graded is thin by the company's own account: a demonstrator running at G-QuAT with AIST ^[7], an SDK and simulator still in development ^[16], and no disclosed workload, access route, user count, price, revenue or paying customer ^[4]. Nor a valuation, ownership breakdown or headcount ^[10]. The nearest thing to an external scorecard is AIST's, which places MoQuren inside its System O project and is integrating OptQC's optics into the ABCI-Q hybrid infrastructure ^[12]. On that path OptQC becomes a subsystem in a national lab's stack, and the first people in a position to say whether the hardware behaves are AIST's researchers rather than anyone paying for machine time.