TD Cowen held its Nvidia buy rating despite Google’s rival AI chips

• TD Cowen reaffirmed Nvidia buy rating Thursday despite Google’s new AI training and inference chip announcement Wednesday
• Super Micro Computer stock fell 10% after losing $1.1 billion to $1.4 billion Oracle contract for Nvidia server racks
• Nvidia announced new AI data center deals in UK, China automotive sector, and cloud infrastructure partnerships same day

• 10% Super Micro Computer share price decline following Oracle contract loss announcement
• $1.1B, $1.4B Value of canceled Nvidia GB300 server rack order from Oracle to Super Micro
• 14 MW AI data center capacity Nscale to build across three BT sites in United Kingdom

Nvidia’s grip on the artificial intelligence hardware market tightened this week even as competitive and legal pressures tested its supply chain.

On Thursday, TD Cowen reaffirmed its buy rating on the chipmaker, citing its continued dominance “in terms of performance and breadth of software ecosystem,” a direct challenge to mounting concerns about Google’s newly announced AI training and inference chips unveiled the previous day.

The same morning Nvidia secured fresh infrastructure partnerships across the United Kingdom, China, and the automotive sector, a coordinated show of strength that underscores the company’s ability to convert technological leadership into concrete commercial deals at scale.

TD Cowen’s Buy Rating Holds Despite Google’s Chip Challenge

The research firm’s Thursday endorsement of Nvidia arrives at a critical juncture for institutional investors reassessing the competitive landscape. Google’s announcement of custom AI chips represents the most credible technical threat Nvidia has faced since its market ascent accelerated in 2023, yet Wall Street’s major research boutiques continue to signal confidence in Nvidia’s staying power.

TD Cowen’s positioning reflects a consensus view that architectural superiority and software maturity remain harder to replicate than raw chip performance specifications.

For institutional investors evaluating exposure to AI infrastructure, the rating reaffirmation carries weight precisely because it comes without qualification or downgrade. The research firm is not hedging Nvidia’s dominance or suggesting a “hold pending competitive clarification”, a stance it could easily justify given the Google announcement.

Instead, the firm doubles down on Nvidia’s defensibility through its software moat, a factor that typically carries more durable competitive advantage than generational chip improvements alone. This messaging suggests that inside institutional decision-making, near-term competitive intensity is being discounted against multi-year infrastructure lock-in effects.

Nvidia Announces 14 Megawatt UK Data Center Buildout with BT and Nscale

On the same day as the TD Cowen rating, Nvidia announced a partnership with British telecommunications company BT and cloud infrastructure firm Nscale to construct AI data center capacity across the United Kingdom. Nscale will build up to 14 megawatts of compute capacity across three existing BT sites, leveraging Nvidia’s full-stack infrastructure.

BT provides the underlying connectivity layer needed to service rising AI workload demand. The project explicitly targets organizations seeking to run sensitive AI systems on domestic infrastructure, sidestepping reliance on foreign-controlled cloud environments, a regulatory and political priority for UK enterprises and government agencies.

This deal structure reveals how Nvidia’s market leadership translates into long-term revenue streams beyond chip sales.

Rather than competing solely on silicon performance, Nvidia captures value across the full stack: its GPUs power the compute layer, its software ecosystem (CUDA, Triton, and related tools) lock in developer productivity, and its partnership model enables regional buildouts that tie infrastructure decisions to broader geopolitical concerns around data sovereignty.

For institutional investors, the deal demonstrates that Nvidia’s defensibility extends beyond chip superiority into systems integration and regulatory alignment.

The UK project addresses a growing appetite among European organizations to develop independent AI infrastructure. Healthcare, energy, finance, and national security applications require local data residency and processing capabilities.

By embedding itself into BT’s existing real estate and telecommunications infrastructure, Nvidia positions its hardware as the default technical standard for this emerging cohort of sovereign AI projects.

This pattern, regional sovereign infrastructure coupled to Nvidia’s hardware stack, is likely to repeat across other European and Asia-Pacific markets, creating durable, region-locked revenue streams.

Desay SV and Nvidia Launch Level 3/4 Autonomous Vehicle Platform in China

Nvidia’s automotive partnerships continue to expand despite a slowing EV market. Working with Chinese autonomous systems developer Desay SV, the companies jointly announced a new intelligent driving solution debuting at the Beijing Auto Show. The system stacks two Nvidia DRIVE AGX Thor chips linked via NVLink interconnect technology, delivering 4,000 FP4 TFLOPS of combined computing power.

The architecture targets the technical requirements for Level 3 and Level 4 autonomous driving, vehicles capable of handling highway and urban environments with minimal or no human intervention under specific driving conditions.

The design choice to use edge-side computing rather than cloud connectivity has institutional importance. By processing all perception, planning, and control decisions locally on vehicle hardware, the system eliminates cloud dependency, a critical requirement for safety-critical autonomous driving applications. Data security improves because raw sensor streams and decision data remain on-vehicle.

Real-time latency requirements for autonomous driving become predictable and verifiable, reducing the risk profile that regulatory bodies will evaluate. For investors tracking Nvidia’s expansion into mobility hardware revenues, this partnership signals that Nvidia has successfully positioned itself as the computing backbone for autonomous vehicle platforms across major automotive markets.

Super Micro Computer’s $1.1 Billion Oracle Contract Cancellation Tests Supply Chain Stability

The morning’s positive news was undercut by supply chain turbulence. Super Micro Computer, a key integrator of Nvidia’s server hardware, suffered a 10% single-day stock decline after reports emerged that Oracle canceled a major order for Nvidia GB300 NVL72 server racks. Research firm Bluefin estimated the canceled contract was worth between $1.1 billion and $1.4 billion.

According to industry sources cited by Bluefin, the cancellation correlates with a lawsuit against Super Micro’s co-founder involving alleged smuggling of AI graphics processors to China.

This represents a concrete disruption in Nvidia’s supply chain, not merely a competitive or regulatory headwind. Super Micro integrates Nvidia’s highest-end GPUs into turnkey data center solutions; losing a customer the size of Oracle removes material revenue from the supply chain partners that make Nvidia’s chips deployable at scale.

The legal and geopolitical dimensions compound the commercial damage. Supply chain concerns around export controls and sanctions compliance increasingly factor into purchasing decisions at major cloud providers and enterprise data center operators.

A contract cancellation tied to alleged sanctions violations signals that procurement officers are applying tighter compliance scrutiny to Nvidia-adjacent suppliers. This may accelerate shifts toward alternative integrators or in-house server design among large hyperscalers, a structural change that would bypass companies like Super Micro entirely.

Bluefin reports that Wistron NeWeb is believed to have assumed the canceled racking business, indicating that alternative suppliers can capture this demand. For institutional investors, the question is whether this represents temporary market-share reallocation or the start of systematic diversification away from single-source suppliers for Nvidia infrastructure.

Institutional investors should monitor whether the Oracle cancellation prompts other large cloud providers and hyperscalers to review their Nvidia server supply chain arrangements over the next two quarters. The scale of the canceled contract and its stated link to compliance issues suggest this is not an isolated incident; rather, it may signal the beginning of a broader rebalancing toward suppliers with stronger export control and sanctions compliance frameworks. Watch for earnings revisions from Super Micro, statements from other Nvidia integrators like Wistron regarding their compliance postures, and any guidance changes from major Nvidia customers about procurement timelines and approved supplier lists.