Appearing on the All-In podcast, Martin Shkreli delivered a blunt verdict on the quantum computing sector that investors should take seriously.
His core argument was simple and direct: “They’ll never beat an NVIDIA machine,” he said, targeting the AI workload that has driven the sector’s sky-high valuations.
Shkreli acknowledged that quantum computers can solve a narrow class of problems faster than any classical machine, particularly factoring the large numbers underpinning modern encryption.
But he was quick to deflate the commercial excitement around that capability, saying: “But to a business person, unless you’re in the business of factoring numbers, there’s not a whole lot to be excited about here.”
Beyond that single trick, he argues the machines offer limited speedups and lack meaningful memory capacity for real-world data workloads.
“There’s no such thing as quantum RAM,” Shkreli said, meaning any problem fed into a quantum processor must arrive with a very small input, ruling out the massive datasets that AI depends on.
AI computing, he contends, is fundamentally a matrix multiplication problem, and GPUs are built precisely for that task at massive scale.
“Multiplying two lists of numbers is what a GPU spends 99.9% of its time doing,” he said, explaining why NVIDIA’s hardware dominates the AI training and inference market.
He also claimed the most important potential customer is unimpressed, noting: “Every time somebody pitches OpenAI, from what I’ve heard about quantum computer, they say, no, we’re not interested,” while acknowledging that information was secondhand.
When a podcast host made the bull case that quantum is where the transistor was before software ecosystems existed, Shkreli dismissed the comparison, asking: “would you go long punch cards and vacuum tubes, right?” and noting those “were stocks at one point.”
The financial reality of the leading quantum companies supports Shkreli’s skepticism about near-term commercial scale in AI applications.
Rigetti Computing has burned cash every single year since fiscal 2021, including $77.2 million in free cash flow in fiscal 2025, against roughly $13 million in revenue.
IonQ’s revenue did grow sharply, rising from $2.1 million in fiscal 2021 to $130 million in fiscal 2025, but the company completed six acquisitions in 2025 alone, including Oxford Ionics, Capella Space, and id Quantique.
IonQ’s latest strategic move reinforces Shkreli’s point about where quantum value actually lives, as the company agreed to acquire cybersecurity firm Elisity in an all-stock deal worth up to $1.3 billion.
The market reacted poorly to that deal, with IonQ shares falling 9% on the announcement, suggesting investors remain uncertain about the company’s expanding acquisition strategy.
NVIDIA’s own $1 billion commitment to U.S. science over five years, which includes quantum computing, actually aligns with Shkreli’s view rather than contradicting it, since the funding targets hybrid quantum-classical systems and NVIDIA’s CUDA-Q platform.
In that model, quantum chips handle specialized edge cases while GPUs continue to carry the dominant AI workload, exactly the limited role Shkreli envisions for the technology.
IonQ currently trades at 19.5x forward sales, its lowest multiple in three years, down sharply from an average of 75.8x over that period and a peak of 215.2x reached in October 2025.
Even at that compressed valuation, 19.5x forward sales remains the kind of premium investors assign to companies expected to capture large, fast-growing markets in the near term.
For quantum stocks to justify their prices, they will need to build real businesses in security, government research, and scientific computing, the smaller domains where quantum’s unique strengths actually apply.
