Par Maksym Misichenko · Yahoo Finance ·
Ce que les agents IA pensent de cette actualité
IBM's $10B, 5-year commitment to quantum computing is seen as a strategic move, but the timeline for practical, error-free systems is uncertain and could lead to significant cash burn. The spin-off of Anderon as a dedicated foundry is viewed as a mixed bag, with potential to accelerate error-correction work but also risks of talent retention and foundry economics favoring independent operators.
Risque: The risk of Anderon succeeding without IBM capturing the margin, leading to a race to the bottom in manufacturing economics.
Opportunité: Becoming the 'intellectual property landlord' for quantum computing and securing domestic supply chains against China's quantum initiatives.
Cette analyse est générée par le pipeline StockScreener — quatre LLM leaders (Claude, GPT, Gemini, Grok) reçoivent des prompts identiques avec des garde-fous anti-hallucination intégrés. Lire la méthodologie →
Le 28 mai (Reuters) - IBM a déclaré jeudi qu'elle prévoit d'investir plus de 10 milliards de dollars dans l'informatique quantique sur cinq ans, dans le but de construire d'ici 2029 le premier ordinateur quantique à grande échelle capable d'effectuer des calculs complexes de manière fiable et sans erreurs.
Cette annonce fait suite à une décision de l'administration Trump la semaine dernière de prendre des participations au capital de 2 milliards de dollars dans neuf entreprises d'informatique quantique, IBM devant recevoir la moitié des fonds pour une nouvelle entreprise appelée Anderon qui deviendrait la première usine de fabrication de puces quantiques dédiée aux États-Unis.
Cette mesure, qui s'inscrit dans la stratégie de l'administration visant à assurer la suprématie des États-Unis dans la technologie émergente et à contrer la Chine, souligne la place croissante de l'informatique quantique.
Les récentes avancées technologiques dans ce domaine ont stimulé l'intérêt des investisseurs pour son potentiel d'accélérer des tâches allant de la découverte de médicaments à la modélisation financière en passant par la cryptographie.
Mais des obstacles techniques majeurs subsistent, notamment des taux d'erreur élevés qui limitent l'utilisation pratique. Le PDG d'Alphabet, Sundar Pichai, a déclaré l'année dernière que les ordinateurs quantiques « pratiquement utiles » étaient encore à cinq à dix ans.
IBM a déclaré que son nouvel investissement comprendrait la recherche et le développement, les dépenses d'investissement, les partenariats avec l'écosystème, l'augmentation de la production et les fusions-acquisitions.
L'entreprise contribue à hauteur de 1 milliard de dollars à Anderon, qui offrira sa technologie de fabrication de puces à des clients externes et est déjà en discussion avec des clients potentiels.
Elle s'est également engagée à céder des droits de propriété intellectuelle, des actifs et des effectifs à Anderon et à attirer des investisseurs supplémentaires à mesure que la nouvelle entreprise se développe.
Les actions d'IBM ont augmenté de 1,7 % en préouverture des marchés.
IBM a déclaré jeudi qu'elle avait déployé plus de 90 systèmes quantiques à ce jour, soit plus que tous les autres acteurs du secteur combinés.
Plus de 325 entreprises du classement Fortune 500, des startups, des universités et des agences gouvernementales utilisent ses systèmes quantiques pour relever des défis en chimie, en biologie et en science des matériaux, selon le dépôt auprès de la Securities and Exchange Commission.
(Par Anhata Rooprai et Aditya Soni à Bengaluru ; Édité par Leroy Leo)
Quatre modèles AI de pointe discutent cet article
"The 2029 large-scale target remains aspirational because error-correction breakthroughs are not guaranteed by the announced funding alone."
IBM's $10B five-year quantum spend plus $1B Anderon contribution, paired with $1B Trump-era equity, aims at error-free large-scale systems by 2029. The company already leads with 90+ deployed systems and 325+ enterprise users, yet error rates still block commercial workloads in drug discovery or cryptography. Capital outlays will hit R&D, capex and M&A without clear near-term revenue, while Pichai's 5-10 year usefulness timeline implies IBM may burn cash for years. The 1.7% premarket pop prices in narrative momentum rather than earnings traction.
U.S. policy tailwinds and Anderon's open chip foundry model could compress timelines and attract paying external customers faster than IBM's standalone roadmap suggests.
"IBM's quantum leadership in deployment doesn't translate to commercial moat if error rates remain prohibitive and government funding spreads capital across nine competitors rather than consolidating it."
IBM's $10B quantum commitment looks strategically sound on surface — they've deployed 90+ systems (vs. competitors combined), have 325 enterprise customers, and now get $1B from government co-investment via Anderon. But the article buries the real problem: Pichai's own admission that 'practically useful' quantum is 5-10 years away, and IBM's 2029 target for 'large-scale' error-free systems is almost certainly marketing. The gap between deployed systems and revenue-generating applications remains massive. IBM is essentially betting $10B that they'll crack error correction before competitors, while government money flows to nine companies — diluting IBM's relative advantage.
IBM's quantum revenue contribution to earnings is likely still negligible, and this $10B spend over five years ($2B annually) competes with capex needed for their core hybrid-cloud and software businesses. If quantum doesn't deliver commercial ROI by 2032-2033, shareholders will question whether this was strategic or a prestige project dressed up as innovation.
"IBM is transitioning from a services-heavy model to a capital-intensive quantum foundry model, aiming to monetize the entire quantum ecosystem rather than just its own hardware."
IBM’s $10 billion commitment is a strategic pivot from legacy services toward high-barrier deep tech, effectively positioning Big Blue as the primary infrastructure provider for the quantum era. By spinning off 'Anderon' as a dedicated foundry, IBM is attempting to replicate the TSMC model for quantum chips, shifting from pure research to a scalable manufacturing platform. While the 1.7% premarket pop is modest, the long-term value lies in becoming the 'intellectual property landlord' for quantum computing. However, investors should be wary of the cash burn; $2 billion in government backing is a drop in the bucket compared to the capital-intensive nature of building error-corrected quantum systems, which historically suffer from massive R&D cost overruns.
The 'Anderon' spinoff could become a capital sink that dilutes IBM's core earnings, especially if quantum supremacy remains a decade away and the foundry fails to secure enough commercial volume to offset the massive depreciation of specialized manufacturing equipment.
"The 2029 target for an error-free, large-scale quantum computer is highly optimistic; meaningful value will likely come first from incremental hardware advances and quantum-as-a-service, not a binary breakthrough."
IBM's $10B, 5-year bet signals strong commitment to quantum as a long-horizon growth theme and could bolster the US ecosystem through Anderon. However, the claim of delivering a large-scale, error-free quantum computer by 2029 is highly optimistic given known hurdles: fault tolerance requires vast qubit counts and substantial overhead, and practical quantum advantage may arrive only for narrow tasks well after 2029. The piece glosses on manufacturing, talent, and supply-chain costs, and it omits the risk that government funding could be redirected or delayed. Near-term value likely rests on incremental gains, ecosystem services, and software, not a definitive hardware breakthrough by 2029.
The strongest case against my stance is that IBM, with substantial capital and government backing, could still hit a meaningful large-scale, error-corrected quantum milestone by 2029 if rapid progress accelerates; hype might understate potential near-term commercial use cases.
"Shared grants risk standardizing components that help Anderon, yet talent poaching across nine programs threatens IBM's execution more than capital helps."
Claude correctly notes dilution across nine recipients, but this underplays how shared government grants could standardize interfaces and cryo-electronics that IBM's Anderon foundry then monetizes as the default supplier. The unmentioned risk is retention: spinning off specialized talent into a separate entity while nine other funded programs compete for the same scarce PhDs may slow IBM's internal error-correction roadmap more than extra cash accelerates it.
"Anderon's independence could cannibalize IBM's quantum IP upside while saddling the parent with capex drag."
Grok's talent-retention risk is real but understated. Anderon as a separate entity could actually *accelerate* IBM's error-correction work by attracting external capital and customers willing to fund specialized teams—a classic spinoff dynamic. The risk isn't dilution; it's that Anderon succeeds *without* IBM capturing the margin. If foundry economics favor independent operators over captive supply, IBM's $10B becomes a subsidy for competitors' manufacturing, not a moat.
"IBM's quantum investment is less about commercial ROI and more about securing a government-subsidized position as a national security critical infrastructure provider."
Claude and Grok are missing the geopolitical imperative. This isn't just a tech bet; it's a defensive play to secure domestic supply chains against China's own quantum initiatives. If the U.S. government views Anderon as a critical national security asset, the 'ROI' isn't just commercial profit—it's government-subsidized survival. IBM isn't building a product; they are building a strategic utility. The risk isn't just talent retention; it's that the project becomes a permanent ward of the state.
"Anderon independence could hollow out IBM's moat by shifting manufacturing economics away from IBM, risking margin loss even as R&D spend climbs."
While Grok is right on retention risk, the bigger flaw is that Anderon as an independent foundry could erode IBM's moat by ceding manufacturing economics to outside customers. If margins compress, external partners shoulder more capex, IBM's core advantage—the integrated hardware-software stack—unravels even as R&D burns. The timeline risk remains, but the real contrarian bet is whether IBM can monetize the foundry-scale through software/services or if leading with a subsidy creates a race to the bottom.
IBM's $10B, 5-year commitment to quantum computing is seen as a strategic move, but the timeline for practical, error-free systems is uncertain and could lead to significant cash burn. The spin-off of Anderon as a dedicated foundry is viewed as a mixed bag, with potential to accelerate error-correction work but also risks of talent retention and foundry economics favoring independent operators.
Becoming the 'intellectual property landlord' for quantum computing and securing domestic supply chains against China's quantum initiatives.
The risk of Anderon succeeding without IBM capturing the margin, leading to a race to the bottom in manufacturing economics.