Qualcomm Goes All-In: The $10B Bet to Crush NVIDIA

Key Points

• Qualcomm is reportedly in talks to acquire RISC-V AI chip startup Tenstorrent for up to $10 billion, accelerating its push into data center infrastructure.
• Qualcomm unveiled its Dragonfly brand at COMPUTEX to target power-constrained hyperscale data centers with energy-efficient AI inference chips.
• JPMorgan assigned QCOM a $265 price target, projecting data center revenue reaching $35 billion by fiscal 2031.

The semiconductor market is experiencing a tectonic shift, and legacy hardware designers are scrambling to secure a seat at the artificial intelligence table.

Many investors are closely tracking Qualcomm Incorporated NASDAQ: QCOM as the company engineers a massive pivot away from cyclical consumer electronics. For decades, Qualcomm built a global empire on mobile smartphone processors, but an aging upgrade cycle demands a new growth engine.

Qualcomm is combining internal product development with targeted acquisitions to penetrate the hyperscale data center ecosystem. By pursuing alternative neural architectures, the firm is positioning itself as a potential low-power, high-efficiency alternative to current AI inference market leaders.

The broader market clearly recognizes the value of this pivot, pushing Qualcomm's stock price up around 30% since the start of 2026. The immediate narrative centers on a massive acquisition target that could further accelerate the balance of power in enterprise computing.

Bypassing the Dealer: Qualcomm Could Drop $10B on RISC-V

A major catalyst driving institutional interest is that Qualcomm is reportedly in talks to acquire AI processor startup Tenstorrent.

Reportedly valued at a $8 billion to $10 billion, this prospective deal would represent a steep premium over Tenstorrent's previous $3.2 billion valuation, reflecting the extreme scarcity of top-tier silicon architecture talent in today's market.

Tenstorrent is led by Jim Keller, a legendary silicon architect whose track record spans fundamental processor designs at nearly every major technology conglomerate over the past two decades.

More importantly, Tenstorrent builds hardware on RISC-V, an open-standard instruction set architecture. This is a highly strategic distinction that investors need to understand. Historically, mobile processors have relied heavily on proprietary ARM architecture, subjecting manufacturers to rigid licensing fees and strict design constraints. Integrating Tenstorrent's RISC-V technology could give Qualcomm more architectural flexibility and reduce reliance on proprietary CPU licensing in certain future products, though it would not eliminate external dependencies across the broader hardware stack.

This maneuver would build on Qualcomm's under-the-radar December 2025 buyout of Ventana Micro Systems. Combining Ventana's high-performance server chiplets with Tenstorrent's neural accelerators would complete a proprietary, non-ARM hardware stack. Instead of retrofitting low-power mobile chips for heavy enterprise workloads, Qualcomm is developing a purpose-built architecture specifically designed to handle intensive data center operations.

Cashing in the Chips: Dragonfly Enters the Server Room

Securing the underlying architecture is only half the battle; deploying hardware effectively in enterprise environments requires a dedicated server platform. At the recent COMPUTEX summit, Qualcomm officially unveiled Dragonfly as a dedicated brand for data center artificial intelligence inference chips.

To fully grasp the market opportunity here, investors must differentiate between training and inference. Training requires massive clusters of graphics processing units that consume vast amounts of electricity to build large language models. Inference is the actual daily application of those models, which includes answering user prompts, executing automated tasks, and processing real-time data streams. Crucially, inference runs continuously.

Hyperscale data centers are currently facing severe power envelope and liquid-cooling constraints. Facilities simply cannot draw enough electricity off the local power grid to run power-hungry training hardware for basic inference tasks. Dragonfly targets this exact physical bottleneck. Positioned heavily for agentic workloads, where models autonomously execute complex, multi-step workflows without constant human prompting, Dragonfly prioritizes power efficiency above all else.

By pairing the Dragonfly server platform with Tenstorrent's specialized hardware accelerators, Qualcomm aims to offer hyperscalers a gigawatt-saving alternative. If physical server rack space and local electricity availability become the primary limiting factors for scaling generative networks, low-power inference hardware provides a distinct, highly defensible competitive moat.

A Royal Flush: Qualcomm's Bulletproof Balance Sheet

The financial metrics firmly support this aggressive expansion phase. Qualcomm shares are currently trading around $220, reflecting a steady year-to-date climb. While Qualcomm recently experienced a 25% technical pullback from peak levels, giving up some decade-high valuation multiples, underlying profitability remains strong. Qualcomm generated $9.20 in trailing 12-month earnings per share (EPS), boasting net margins of 22.31% and an impressive 42.11% return on equity.

Legacy markets are undeniably contracting. Core handset revenue declined 13% year over year in the second quarter of fiscal 2026, pressured by inflation in memory components and suppressed production volumes in key Asian markets. The diversification strategy is already bearing fruit, offsetting these headwinds. The company's automotive revenue surged 38% year over year, surpassing a $5 billion annualized run rate.

Wall Street is actively adjusting financial models to account for the shifting revenue base. JPMorgan analysts recently placed Qualcomm on a Positive Catalyst Watch, raising the price target to $265. Their aggressive modeling projects Qualcomm data center revenue scaling rapidly, hitting $3 billion by fiscal 2027 and accelerating to $35 billion by fiscal 2031.

The River Card: Securing Your Stake in Qualcomm

Qualcomm is signaling immense balance sheet confidence ahead of these capital-intensive integrations. The board of directors recently raised the quarterly dividend to 92 cents per share.

This dividend hike operates concurrently with a massive $20 billion share repurchase program authorized in March 2026. This buyback program allows Qualcomm to retire up to 14.5% of outstanding stock, providing a strong structural floor during broader market rotations.

Institutional focus is squarely fixed on the upcoming June 24 Investor Day. Markets anticipate detailed roadmaps outlining the potential Tenstorrent integration, the broader Dragonfly rollout, and updated margin guidance.

Transitioning from a cyclical handset supplier to a foundational enterprise infrastructure provider carries execution risks, particularly when challenging entrenched industry incumbents. The strategic pivot aligns perfectly with the most pressing pain point in the global technology sector: the need for affordable, energy-efficient computing power.

Investors seeking exposure to the next phase of the digital infrastructure buildout may want to add Qualcomm to their watchlists as it builds out a new data center footprint and executes an aggressive acquisition strategy.