THE APEX TIMES
Martin Shkreli backs NVIDIA on AI, arguing quantum computers will not outpace GPUs
On the All-In podcast, entrepreneur Martin Shkreli said quantum computing is unlikely to “beat an NVIDIA machine,” framing GPUs as the near-term engine for artificial intelligence workloads.
Martin Shkreli used a discussion about quantum computing to make a blunt case for today’s dominant hardware in artificial intelligence: GPUs from NVIDIA. In remarks published this week, the activist-turned-entrepreneur argued that quantum computers will not displace GPU-based systems for AI in the way investors often anticipate.
The comments were shared in a segment associated with the All-In podcast. Shkreli’s central point was comparative performance, not long-term theoretical potential. He suggested that even as quantum capabilities mature, they will not produce results that surpass what current NVIDIA-equipped machines can deliver for AI applications.
Shkreli also made a demand-side argument about how resources are allocated in computing today. By his count, GPUs spend roughly 99% of the time working on AI-relevant computation, implying that the practical roadmap for most companies will continue to be oriented around accelerators optimized for large-scale training and inference.
In the same discussion, Shkreli drew a line between what quantum computing is good at and what modern AI infrastructure is built to do. Quantum processors, he indicated, are not on a path to replace the current approach to AI hardware. Instead, he framed GPUs as the platform where performance and software ecosystems have already converged, lowering the risk for companies deploying AI at scale.
For NVIDIA, such public endorsements speak to a broader narrative the company has long benefited from, namely that GPUs are not just faster chips, but the core of an end-to-end AI system. NVIDIA’s product strategy in the data center has emphasized that buyers want integrated stacks, including high-throughput computing and software tooling that helps developers run increasingly demanding AI models.
That framing matters because the AI market has rewarded both raw compute and the availability of software libraries that make high-performance hardware usable. GPU platforms are supported by widely adopted programming models, developer tools, and performance-optimized libraries, factors that can make switching costly even when alternative technologies look promising in theory.
Still, Shkreli’s remarks do not resolve a key uncertainty for the market: when, or whether, quantum computing could become commercially relevant for AI workloads. His comments were comparative and time-horizon oriented, but they did not provide specific timelines, benchmark results, or workload details that would let investors translate his view into a measurable forecast.
What to watch next is how the market continues to weigh hardware optionality. As quantum research progresses, investors will likely look for clearer evidence of performance advantages on AI-relevant tasks and for signs that quantum ecosystems are converging with production-scale software. In the meantime, public assertions like Shkreli’s reinforce the prevailing belief that NVIDIA-style GPU acceleration remains the default path for AI deployment.
Why It Matters
- The comments underline investor and public focus on whether quantum computing can realistically compete with GPU acceleration for mainstream AI.
- They reinforce NVIDIA’s positioning as a central vendor for AI infrastructure, not only as a chip maker but as a platform.
- Because Shkreli did not cite specific quantum-AI benchmarks, the market still lacks concrete performance evidence to overturn the current GPU-led approach.
- As quantum efforts advance, the comparison will likely shift from general claims to workload-specific results that can be validated in real deployments.
Sources
Key Facts
- Martin Shkreli made the case that quantum computers will not “beat an NVIDIA machine” for AI.
- His comments were published in connection with the All-In podcast discussion.
- Shkreli argued that GPUs are where most AI computation time is currently spent, estimating roughly 99%.
- The remarks centered on hardware comparison and near-term practicality rather than providing benchmarks or timelines.
- NVIDIA is the named beneficiary in Shkreli’s argument, with GPUs positioned as the enduring AI platform.
Technology Related
Meta’s Fiber Push and Corning’s Pricing Power: AI Data Center Demand Meets Higher Optical Fiber Costs
A forecast of sharply higher prices for specialized AI data-center fiber could change the near-term economics for Corning, Meta’s reported fiber partner, even as the broader link between pricing and company earnings remains unclear.
NVIDIA and Jülich researchers report a record simulation tied to fault-tolerant quantum computing on Europe’s exascale system
The simulation, conducted on JUPITER at the Jülich Supercomputing Centre, aims to model some of the most demanding calculations needed for fault-tolerant quantum computing.
Jensen Huang says NVIDIA wants AI “token” costs to keep dropping, aiming for 30 to 40x per year
The NVIDIA CEO tied his push for cheaper AI computation to a simple dynamic: if tokens get much less expensive while demand keeps rising, overall usage can grow faster than prices fall.
Microsoft unveils Maia 300 AI chip, but one analyst says it is “nowhere close” to NVIDIA or Google
Microsoft says it is moving ahead with its in-house Maia AI chips and targets large-scale deployments beginning next calendar year. Still, an analyst at Bloomberg Intelligence argues Microsoft’s designs are not yet competing at the same level as the leaders in AI accelerator hardware.
Morgan Stanley frames AWS growth as a path to higher Amazon earnings, lifting AMZN optimism
A Wall Street note is fueling bullish sentiment for Amazon shares, with expectations that AWS can grow about 47% in 2027 and support earnings power around $16.50 by 2028, according to the report cited by market media.
Apple confirms Oct. 28 “Pro and Tell” event, with a first touchscreen MacBook Pro rumored
The company said it will hold an October 28 product event dubbed “Pro and Tell,” as market chatter points to a touchscreen MacBook Pro and investors watch whether recent early demand for a foldable iPhone “Duo” can sustain momentum.
Munro Partners flags Nvidia valuation gap as its Growth Fund trails for Q3 2026
A new third-quarter 2026 letter from Munro Global Growth Fund argues Nvidia’s share price is not keeping pace with the company’s earnings momentum, even as the fund posted a net decline during the quarter.
Meta launches “Muse Wonder” AI models, staking a bid for top-tier performance in generative media
Meta said it rolled out Muse Wonder and Muse Wonder Full on Oct. 12, describing them as its most capable AI models yet. Investors and analysts are framing the move as a step toward challenging leaders in AI image and creative tooling, even as questions remain about pricing, availability, and how quickly developers will adopt the new models.
Amazon plans to back 12 small nuclear reactors at former Indiana coal plant through X-energy deal
The cloud and retail giant is reported to be working with X-energy on a nuclear deployment designed to supply close to 1 gigawatt of electricity from a retired coal site in Indiana.
Amazon shares climb after Morgan Stanley highlights AWS “runway” and lifts target
Amazon.com’s stock rose more than 5% on Monday after Morgan Stanley reiterated the shares as a top pick and raised its price target, pointing to what it called an “under-appreciated” runway for Amazon Web Services.