Overview

On July 2, 2026, Nvidia announced a landmark partnership program that allows fast-growing AI startups to swap access to GPU compute power for a share of their future revenue [1]. Under the program, cloud-based AI firms, model builders, and other enterprises receive token credits to power their development, and in return share both product and cloud revenue with Nvidia [1]. The initiative, detailed in an official Nvidia blog post titled "NVIDIA Unlocks AI Compute at Scale, Inviting Partners to Power the AI Infrastructure Buildout," represents a structural shift in how AI infrastructure is financed and accessed [4]. Two initial infrastructure partners were named: Australia-based Sharon AI, which will deploy up to 40,000 Nvidia GPUs, and Singapore-based Firmus Technologies, which is building a 360-megawatt data center campus in Batam, Indonesia, expected to house up to 170,000 Nvidia GPUs [1][2][3]. Together, these two partners alone offer potential access to more than 200,000 Nvidia GPUs for AI startups [1].

The program arrives at a moment of extraordinary tension in the AI ecosystem. GPUs have been likened to oil and are reportedly tied to futures contracts as users grapple with wild fluctuations in cost and availability [1]. AI firms have increasingly turned to revenue- and equity-sharing agreements with chipmakers to circumvent liquidity issues; OpenAI, for example, has inked similar arrangements with Amazon and AMD [1]. Nvidia's move positions the company not merely as a hardware supplier but as a financier and gatekeeper of AI compute capacity, with profound implications for startups, competitors, and the broader AI infrastructure landscape.

---

1. Program Terms & Structure

1.1 The Revenue-Sharing Model

The program operates through Nvidia's existing DSX program (Data Center Infrastructure for AI), which partners with data center operators to deploy GPU infrastructure on a revenue-sharing basis rather than requiring upfront purchases [2]. This economic alignment model is designed to make advanced AI infrastructure accessible to a broader range of customers who lack the balance-sheet strength of hyperscalers.

Under the specific terms disclosed, Nvidia earns standard product revenue from the initial GPU sale plus a share of the ongoing cloud revenue generated by the partner [2]. AI startups, in turn, receive token credits to power their development and share both product and cloud revenue with Nvidia [1]. Nvidia acts as an intermediary, giving startups direct access to full-stack computing powered by its GPUs [1]. The precise percentage of revenue shared, duration of revenue-sharing obligations, and any caps on total payments have not been publicly disclosed as of July 2, 2026. However, the Firmus Technologies partnership is structured as an eight-year agreement running through 2034, suggesting that Nvidia is taking a long-duration view on these revenue streams [2][3].

The program is explicitly designed to address a structural inequity in AI infrastructure access. As Firmus Co-CEO Tim Rosenfield explained: "We have worked to figure out how to close the gap between the cost benefits that the large guys have access to, which they do because they have great credit ratings, and the guys that are up and comers. This is actually a really material way to level the playing field a little bit to give the next a chance to compete with the big guys" [2].

1.2 Initial Infrastructure Partners

Sharon AI (Australia-based) will deploy up to 40,000 Nvidia GPUs under the program [1]. Details on Sharon AI's specific deployment timeline and facility specifications remain limited in the initial announcement.

Firmus Technologies (Singapore-based) represents the larger and more detailed of the two initial partnerships. Key facts about the Firmus deployment include:

• A 360-megawatt data center campus in Batam, Indonesia, an island near Singapore [1][2][3]
• Capacity for up to 170,000 Nvidia GPUs [1][2][3]
• Construction underway with Singapore-based DayOne, with operations expected by Q1 2027 [2][3]
• Deployment of Nvidia AI accelerators across the Grace-Blackwell, Vera-Rubin, and Vera platforms through 2027–2028 [2]
• One of the largest AI infrastructure deployments in the Asia-Pacific region [2]
• A multi-tenant facility designed for AI-native customers, distinct from Firmus's Australian projects that target hyperscalers [2]
• Firmus expects $25–30 billion from committed offtake agreements in the first six years of the partnership [2][3]
• The partnership is structured as an eight-year agreement (through 2034) [2][3]

Firmus Technologies was originally founded in 2019 as a cryptocurrency mining operation before pivoting to AI infrastructure [2]. The company was valued at $5.5 billion in an April 2026 investment round backed by Nvidia and is preparing for an IPO in 2026 [2]. Firmus uses Nvidia's reference architecture with its proprietary liquid-cooled HyperCube design [2]. The company is also expanding across Australia through Project Southgate, which involves approximately 18,400 Nvidia GPUs in Melbourne, and plans to invest over $70 billion in AI infrastructure across Australia, backed by $1.35 billion in equity and a $10 billion debt facility [5]. Firmus also secured a landmark 12-year wholesale energy supply agreement with Gunvor Group for 600 MW of firm electricity to support Project Southgate, which will underpin the development of 1.2 GW of new renewable generation and 1.5 GWh of new battery storage by 2032 [6].

1.3 Target Startups and Selection Criteria

The program targets fast-growing AI startups broadly, including cloud-based AI firms, model builders, and other enterprises needing compute power [1]. The Batam facility is explicitly described as a multi-tenant project designed for AI-native customers, as opposed to hyperscalers [2]. The program aims to serve "AI-native companies" that need "access to scalable, energy and cost-efficient compute infrastructure to compete globally" [2].

As of the July 2, 2026 announcement, Nvidia has not publicly disclosed formal evaluation criteria or a detailed selection process for startup participants. The program is described as being for "fast-growing" AI startups, and the initial partners (Sharon AI and Firmus Technologies) are both AI infrastructure companies that build and operate data centers to provide compute to end customers, rather than end-user startups themselves. This suggests a two-tier structure: Nvidia partners with data center operators through the DSX program, and those operators in turn provide compute access to AI startups under the revenue-sharing framework.

1.4 Compute Allocation

No formal minimum or maximum compute allocation limits have been specified for individual startup participants. However, the two initial partner deployments provide reference points for the scale of the program: Sharon AI at 40,000 GPUs and Firmus Technologies at 170,000 GPUs, for a combined initial capacity exceeding 200,000 GPUs [1]. The token credit system suggests that individual startup allocations will be flexible and tailored to specific needs, though the mechanics of how credits are allocated, tracked, and reconciled against revenue have not been publicly detailed.

---

2. Strategic Rationale for Nvidia

2.1 Why Now: The AI Infrastructure Buildout

Nvidia CEO Jensen Huang has described the current moment as one in which "the buildout of AI factories — the largest infrastructure expansion in human history — is accelerating at extraordinary speed" [7]. The revenue-sharing program is a direct mechanism to accelerate this buildout by removing the financing bottleneck for startups that lack the credit ratings of hyperscalers.

Several converging factors explain the timing of this program in mid-2026:

GPU Scarcity as a Strategic Lever. Access to GPU compute has become the single most critical bottleneck for AI startups. GPUs are compared to oil and reportedly tied to futures contracts as users grapple with cost fluctuations and availability issues [1]. By acting as an intermediary that provides direct access to full-stack computing, Nvidia positions itself as the essential gatekeeper of AI compute capacity, deepening the dependency of the AI startup ecosystem on its hardware and software stack.

The Hyperscaler Spending Surge. Amazon, Microsoft, Alphabet, and Meta are together on track to spend approximately $725 billion on capital projects in 2026 — up about 77% from 2025 — with most of it pointed at AI infrastructure [7]. This spending is increasingly debt-funded, and free cash flow is under pressure for these customers [7]. Nvidia's program offers an alternative path for startups that cannot compete with hyperscaler budgets.

Sovereign AI and Government Markets. Nvidia is also pursuing sovereign AI deployments. On June 29, 2026, Palantir Technologies announced a strategic initiative with Nvidia to deliver an intelligent engine for deploying Nvidia AI and Nemotron open models in sovereign environments, primarily targeting U.S. government agencies and critical infrastructure [25]. Jensen Huang stated: "Open source AI is foundational to national security, public safety and U.S. technology leadership" [25]. The revenue-sharing program extends this logic to the private startup ecosystem.

Nvidia's Own Capital Strategy. Nvidia announced plans in June 2026 to raise at least $20 billion in debt for general corporate purposes, including refinancing [1]. The company has $119 billion in supply commitments and an $80 billion buyback authorization [9]. Following a 2,400% dividend payout increase, Nvidia is on track to become a Dividend Aristocrat, backed by unprecedented free cash flow [29]. The revenue-sharing program represents a deployment of Nvidia's financial strength to capture a new category of recurring revenue.

2.2 Financial Implications

Revenue Recognition and Business Model Transformation. Under the revenue-sharing model, Nvidia earns standard product revenue from the initial GPU sale plus a share of the cloud revenue generated by the partner [2]. This creates a recurring revenue stream on top of the upfront hardware sale, transforming Nvidia's business model from purely transactional hardware sales to a hybrid model with ongoing participation in the AI cloud services market. The Firmus deal alone is expected to generate $25–30 billion from committed offtake agreements in the first six years [2][3], with Nvidia taking a share of that cloud revenue in addition to the hardware revenue.

Current Financial Performance. In fiscal Q1 2027 (ended April 26, 2026), Nvidia's revenue rose 85% year over year to $81.6 billion, with AI-focused data center revenue up 92% to $75.2 billion [7]. Management guided for approximately $91 billion in fiscal Q2 [7]. Nvidia reported a 75% non-GAAP gross margin and $48.55 billion in quarterly free cash flow [9].

Margin Considerations. Nvidia's 75% gross margin is a key competitive advantage [9]. The revenue-sharing model could put modest pressure on gross margins in the short term as Nvidia takes on more financing risk, but it could also expand operating margins over time by generating high-margin recurring revenue streams from cloud services. The bear case for Nvidia is that its ~75% gross margin could narrow if competition from in-house chips (Alphabet, Amazon, Microsoft, Meta) and AMD's accelerators erode pricing power [7].

Stock Performance Context. Nvidia's stock has fallen approximately 18% from its mid-May 2026 high to around $193 as of late June 2026, and is up only about 3% year-to-date [7][26]. The stock trades at about 30 times earnings, well below the 40-plus multiple it carried for much of the past two years [7]. Analysts estimate that if the AI build-out continues and margins hold, the stock could compound at a high-single-digit to low-double-digit annual rate, reaching the high-$200s to low-$300s by 2030 [7]. If AI spending peaks within one to two years and competition softens pricing, the stock could stagnate [7].

2.3 Comparison to Existing Cloud Partnerships

Nvidia's existing cloud partnerships provide important context for understanding the new program:

CoreWeave is one of Nvidia's most important neocloud partners, having attracted more than $99 billion in contracts [8]. CoreWeave was the first cloud provider to integrate Nvidia's Vera Rubin NVL72 platform [8]. In Q1 2026, CoreWeave reported revenue of almost $2.1 billion (up 112% year-over-year), though growth slowed from 167% in 2025 [8]. However, CoreWeave posted a $740 million loss and carries nearly $25 billion in debt against $4.8 billion in book value [8]. The stock is down over 40% in the past year but up more than 40% year-to-date, trading at a price-to-sales ratio of 8 [8].

Together AI, an AI neocloud founded in 2022, raised an $800 million Series C at an $8.3 billion valuation on July 1, 2026, with participation from Nvidia [11]. Together AI claims annual bookings of over $1.15 billion as of its last quarter, driven by increasing adoption of open-source AI models via neocloud providers [11].

Bitdeer AI, a preferred Nvidia Cloud Partner, won the "AI Cloud Platform of the Year" award in the 2026 AI Breakthrough Awards [17]. The company employs a vertically integrated "AI Factory" model, owning and operating its own high-performance data center infrastructure [17].

The key distinction of the new compute-for-revenue share program is that it extends the DSX revenue-sharing model — previously used for large infrastructure partnerships like Firmus — to a broader set of AI startups through a standardized token credit system. Unlike the bespoke partnerships with CoreWeave or Together AI, the new program is designed to be scalable and accessible to a wide range of AI-native companies.

---

3. Impact on AI Infrastructure Financing

3.1 Comparison to Traditional Financing Models

Nvidia's revenue-sharing program represents a structural innovation in AI infrastructure financing that differs materially from existing models:

Cloud Credits. Hyperscalers (AWS, Azure, GCP) have long offered startup credit programs that provide prepaid credits for cloud services. These programs typically require startups to be associated with an accelerator, venture fund, or partner network, and credits are finite and expire. Nvidia's program ties compute access directly to future revenue rather than upfront credit allocations, creating an ongoing alignment of incentives between Nvidia and the startup [1].

Venture Debt. Traditional venture debt requires startups to have sufficient revenue or collateral to service fixed debt payments. Nvidia's program eliminates this requirement by taking a share of future revenue instead of fixed payments, making it accessible to pre-revenue or early-revenue AI startups that would not qualify for venture debt [1][2].

Traditional Equipment Leasing. Standard GPU leasing requires upfront payments or strong credit ratings — precisely the barrier that Firmus Co-CEO Tim Rosenfield identified when he said the program aims to "close the gap between the cost benefits that the large guys have access to... and the guys that are up and comers" [2].

Equity-for-Compute. Some AI chipmakers and cloud providers have accepted equity in startups in exchange for compute access. Nvidia's revenue-share model is distinct in that it does not dilute startup equity; instead, it takes a share of operating revenue, which may be more attractive to founders and existing investors.

The program effectively positions Nvidia as a venture-capital-like financier for AI startups, but with a revenue-share structure rather than an equity structure. This aligns Nvidia's incentives with startup success — Nvidia only benefits if the startup generates revenue — while avoiding the dilution and governance implications of equity investments.

3.2 The Broader Debt-Fueled AI Buildout

The Nvidia program must be understood within the context of an AI infrastructure buildout that is increasingly financed with debt. Amazon, Microsoft, Alphabet, and Meta are together on track to spend approximately $725 billion on capital projects in 2026 — up about 77% from 2025 [7]. AI-related debt now approaches 15% of U.S. investment-grade issuance [22]. Morgan Stanley projects 2026 investment-grade issuance could exceed $2 trillion for the first time [22].

Banks are developing new debt-selling strategies to finance this buildout. Hyperscalers have issued $60 billion in multi-currency bonds over the past 12 months, with Amazon raising €14.5 billion in March 2026 — the largest-ever euro corporate bond deal — while Alphabet set records in yen, Canadian dollar, Swiss franc, and sterling [22]. Bankers are also structuring new deals for AI startups and data center operators, such as notes backed by pre-arranged data center leases (e.g., Stingray Compute's $810 million note, nine times oversubscribed, backed by an Amazon lease) [22].

Oracle's aggressive debt-funded AI expansion has created a $638 billion AI backlog, but over half of that backlog is tied to a single customer, OpenAI, raising significant concentration risk [16]. Unlike cash-rich hyperscalers, Oracle relies heavily on borrowing to finance its cloud infrastructure, and the stock has fallen 57% from its 52-week high [16].

Against this backdrop, Nvidia's revenue-sharing model offers an alternative financing mechanism that does not add to startup debt burdens or require the credit ratings that traditional debt markets demand. It is a recognition that the existing financial infrastructure — bank debt, corporate bonds, venture capital — is insufficient to meet the compute needs of the AI startup ecosystem.

---

4. Competitive Implications

4.1 Cloud Providers and GPU-as-a-Service Companies

Meta Platforms: A New Entrant. On July 1, 2026 — one day before Nvidia's announcement — Bloomberg reported that Meta is developing plans for a cloud infrastructure business called "Meta Compute" to sell access to AI computing power and models, positioning itself to compete with AWS, Azure, and Google Cloud [12][13]. The initiative is led by Santosh Janardhan (Meta's head of infrastructure), Daniel Gross (Meta Superintelligence Labs), and Meta President Dina Powell McCormick [13]. Two approaches are being considered: selling access to AI models hosted on Meta's infrastructure (similar to AWS Bedrock) and selling "raw" computing capacity (similar to neocloud providers like CoreWeave) [13]. CEO Mark Zuckerberg signaled openness to selling excess compute during a May 2026 shareholder call, stating it is "definitely on the table" and that "almost every week, there are different companies that come to us from the outside asking us to both stand up an API service or asking if they have compute that they could buy from us at some premium to what we've bought it at" [13]. Following the news, Meta shares jumped 9.3%, while CoreWeave fell as much as 14% and Nebius Group fell as much as 17% [12]. Nvidia shares fell about 2% on the news [12].

Google and TPUs. Alphabet is leveraging its custom tensor processing units (TPUs) as a key competitive advantage. TPUs are application-specific integrated circuits co-designed with Broadcom, optimized for machine learning training and inference, offering better performance per dollar and 20–40% lower energy consumption than Nvidia's GPUs [14]. This cost efficiency allows Google to charge 20–30% less for excess compute capacity, attracting AI unicorns like Anthropic and Meta to its cloud business [14]. Google Cloud revenue is projected to surge approximately 64% to $96 billion in 2026, with continued growth above 50% in 2027 [14]. Google's eighth-generation TPUs are up to 3× faster for training, offer 80% better performance per dollar, and can cluster over 1 million chips [14]. Google has also formed a $5 billion TPU cloud venture with Blackstone [14]. Analysts at Citizens JMP estimate TPU-related infrastructure revenue could reach $3 billion in 2026 and $25 billion in 2027 [14].

SoftBank and its telecom unit announced on July 2, 2026 that they will also rent computing resources to U.S. firms [15].

CoreWeave's Vulnerability. CoreWeave junk bonds slid further on July 2, 2026 as investors questioned the AI boom, with moves by Meta and SoftBank exposing fault lines in the growing pile of AI debt [15]. The entry of Meta — a hyperscaler with 20 GW of capacity worldwide and plans to add 14 GW — directly threatens neocloud providers like CoreWeave that have built their businesses on Nvidia partnerships [13].

4.2 AI Chip Rivals

AMD. AMD has seen its stock surge over 140% year-to-date as of late June 2026 [18]. Trailing 12-month revenue reached $34.64 billion (up 34.34% YoY), with net income of $4.33 billion (up 164.17% YoY) [18]. Strategic partnerships include Meta (6 GW of GPUs) and OpenAI (6 GW of GPUs) [18]. AMD has committed up to £2 billion for AI innovation in the UK and announced over $10 billion in Taiwan ecosystem investments for AI infrastructure [18]. UBS raised AMD's price target to $670 from $455, citing agentic AI-driven CPU demand, and Gartner named AMD the "company to beat for enterprise AI server CPUs" [18].

Cerebras Systems. Cerebras reported Q1 GAAP revenue of $193.4 million (up 94% YoY), with cloud services growing 178% [9]. The company secured a multi-year, $20+ billion OpenAI inference deal covering 750 megawatts [9]. Independent benchmarks show Cerebras' wafer-scale design delivering a 21x speed advantage over Nvidia for latency-sensitive, low-batch inference [9]. However, management guided full-year operating margins to negative 28% to negative 32% [9]. The critical vulnerability: every major LLM framework and enterprise developer stack is natively optimized for Nvidia architecture out of the box, while Cerebras requires specialized compilation and custom engineering support [9].

Etched. AI chip startup Etched achieved a $5 billion valuation and booked $1 billion in contract orders for its "frontier inference clusters" — full systems combining custom chips, racks, and software designed to run AI inference faster and more efficiently [10]. The company's chip was successfully manufactured by TSMC earlier in 2026 [10]. Founded in 2022 by Harvard dropouts and Thiel fellows, Etched has raised $800 million to date, with notable investors including Andrej Karpathy, Geoffrey Hinton, Fei-Fei Li, Stanley Druckenmiller, and Peter Thiel [10].

Groq raised $650 million [10].

Custom Chips from AI Labs. OpenAI is collaborating with Broadcom on a specialized chip called "Jalapeño" designed to accelerate AI model inference and optimize energy consumption [9][19]. Anthropic is in talks with Samsung to jointly produce a new generation of custom AI chips, aiming to reduce dependence on Nvidia [20]. Major technology companies including OpenAI, SpaceX, and Google are developing their own specialized chips to reduce dependence on Nvidia, a trend about diversifying risk rather than completely severing ties [19].

China and Huawei. Nvidia's AI chip sales in China have stalled as domestic rival Huawei gains market share. Nvidia's share of China's AI chip market fell from approximately 95% before export bans to roughly 40% in 2025, matching Huawei, and Bernstein predicts Nvidia will drop to approximately 8% in 2026 while Huawei rises to approximately 50% [21]. Huawei's Ascend 950 series is considered comparable to Nvidia's H200 [21].

4.3 Lock-In Effects and the CUDA Moat

Nvidia's primary competitive advantage is its CUDA software platform. As one analysis noted: "Every major LLM framework and enterprise developer stack is natively optimized for Nvidia architecture out of the box, while Cerebras requires specialized compilation and custom engineering support for anything off the well-trodden path" [9]. Jensen Huang has stated: "NVIDIA is the only platform that runs in every cloud, powers every frontier and open source model, and scales everywhere AI is produced" [9].

The revenue-sharing program could significantly entrench this dominance. By providing token credits that can only be used on Nvidia-powered infrastructure, the program creates a natural path dependency where startups build their models and workflows on CUDA-optimized frameworks from the earliest stages of development. The switching costs — retraining models, rewriting inference pipelines, re-optimizing for different hardware — become prohibitively high once a startup has built its entire product on Nvidia's stack.

This lock-in effect is reinforced by Nvidia's $119 billion in supply commitments and $80 billion buyback authorization, signaling that management is doubling down on the platform [9]. The company's 75% non-GAAP gross margin, $48.55 billion in quarterly free cash flow, and massive developer install base are described as "tough to dislodge in a single product cycle" [9]. However, as one analyst noted: "If CUDA defections spread beyond OpenAI, my view changes" [9].

---

5. Broader AI Ecosystem Effects

5.1 Startup Formation and Innovation

The Nvidia program directly addresses the critical scarcity of GPU compute power that has been a major barrier to AI startup formation. By providing token credits in exchange for future revenue, Nvidia enables startups to access compute without depleting their cash reserves or giving up equity. This could accelerate AI startup formation by removing the compute bottleneck that has forced many startups to either partner with hyperscalers or raise large amounts of venture capital just to secure GPU access [1][2].

The program could also reshape AI startup business models by making compute costs variable rather than fixed. Startups pay a share of future revenue rather than upfront costs, aligning Nvidia's incentives with startup success. This is similar to how OpenAI has inked revenue- and equity-sharing deals with Amazon and AMD [1].

However, the program also creates potential dependencies. Startups that build on Nvidia's platform through this program may find it difficult to switch to alternative hardware later, given the CUDA software lock-in. This could reduce long-term innovation by limiting the diversity of hardware platforms that startups explore. The broader AI chip startup ecosystem is booming — Etched, Cerebras, and Groq have all raised significant capital — and the current funding environment is described as "a different planet" compared to 2023, with investors aggressively chasing AI chip technology [10].

5.2 Antitrust and Regulatory Concerns

FTC Independence Ruling. On June 29, 2026, the U.S. Supreme Court ruled in Trump v. Slaughter that the President has the constitutional authority to fire Federal Trade Commission (FTC) commissioners without cause, overruling the 90-year-old precedent Humphrey's Executor v. U.S. (1935) [23][36]. Chief Justice Roberts, writing for the majority, held that the FTC Act's for-cause removal protection violates the separation of powers because the FTC performs "execution of the law" — the President's constitutional role [23]. Justice Sotomayor, joined by Justices Kagan and Jackson, dissented, arguing that the decision "upends" the separation of powers and will transform independent commissions into purely executive agencies [23]. This ruling could significantly affect how antitrust enforcement against Nvidia is conducted, as the FTC is now more directly under presidential control.

Government Equity in AI Companies. OpenAI has proposed giving the U.S. government a 5% equity stake in the company, valued at roughly $42.6 billion based on OpenAI's $852 billion March 2026 valuation [24]. The proposal follows precedent: the U.S. already holds a 10% stake in Intel and takes a cut of Nvidia and AMD's China AI chip sales (15–25% of revenue) [24]. Public Knowledge's Nat Purser warned the arrangement could make government "less willing to impose, or enforce, safety rules because doing so could reduce the value of its own investment" [24]. The article notes that "government equity in frontier AI labs is a structurally different regulatory posture than subsidies or export controls, and it is arriving piecemeal… rather than through a single deliberate policy" [24].

Export Controls and Chip Smuggling. Taiwanese authorities raided Super Micro Computer offices and affiliated sites as part of an investigation into alleged smuggling of Nvidia AI chips to China, in violation of U.S. export controls [33]. SMCI stock fell 8.10% on the news [33]. The probe forms part of a broader enforcement effort to limit diversions of high-end AI components into restricted markets.

Nvidia's China Market Erosion. Nvidia's AI chip sales in China are stalling as local chipmakers like Huawei take the lead, and the U.S. government takes a revenue cut of Nvidia and AMD's China AI chip sales [21][24]. This dynamic complicates the global competitive landscape and adds regulatory risk to Nvidia's international revenue streams.

5.3 Investor and VC Perspectives

Nvidia's Stock and Valuation. Nvidia stock has fallen about 18% from its mid-May 2026 high to approximately $193 as of late June 2026, and is up only about 3% year-to-date [7][26]. The stock trades at about 30 times earnings, well below the 40-plus multiple it carried for much of the past two years [7]. D.A. Davidson's Gil Luria commented that "Micron and Nvidia are trading like the AI cycle is peaking now" [34]. Applied Digital CEO Wes Cummins predicted "pretty significant delays" for the AI industry through 2026 and 2027 [34].

Bull Case. In fiscal Q1 2027, Nvidia's revenue rose 85% year over year to $81.6 billion, with data center revenue up 92% to $75.2 billion [7]. Management guided for approximately $91 billion in fiscal Q2 [7]. Nvidia's next-generation Vera Rubin platform is due in H2 2026 [7]. If the build-out continues and margins hold, the stock could compound at a high-single-digit to low-double-digit annual rate, reaching the high-$200s to low-$300s by 2030 [7].

Bear Case. The massive AI infrastructure spending is increasingly funded with debt, and free cash flow is under pressure for Nvidia's customers [7]. Competition is rising: Nvidia's biggest customers (Alphabet, Amazon, Microsoft, Meta) are designing in-house chips, and AMD is pushing its own accelerators [7]. Nvidia's ~75% gross margin could erode if pricing power weakens [7]. If AI spending peaks within a year or two and competition softens pricing, the stock could spend years going nowhere even as revenue grows [7].

VC Investment in AI Chips. The funding environment for AI chip startups is booming. Etched raised $800 million at a $5 billion valuation [10]. Cerebras had a breakout IPO [9]. Groq raised $650 million [10]. Investors are chasing everything AI-related, especially chip technology that speeds up inference. Notable investors in Etched include VentureTech Alliance, Jane Street, Hudson River Trading, Two Sigma, Ribbit Capital, and angel investors Andrej Karpathy, Geoffrey Hinton, Fei-Fei Li, Arthur Mensch, and Scott Wu, plus billionaires Stanley Druckenmiller and Peter Thiel [10].

Nvidia's Broader Investment Strategy. Nvidia itself is a major investor across the AI and quantum computing ecosystem. The company deployed $1.6 billion in quantum computing investments, alongside BlackRock ($1.7 billion) and Temasek [30]. VC investment in quantum computing hit $3.9 billion across 125 deals in 2025 [30]. Nvidia will also embed its ARC-Pro processors into Nokia's 5G equipment to enable AI inferencing from cell towers, supporting Nvidia's CUDA platform at the edge [8]. Nokia's stock has surged approximately 170% over the past year [8].

The Revenue-Share Program as a Strategic Hedge. The compute-for-revenue share program serves multiple strategic purposes for Nvidia: it expands the addressable market to capital-constrained AI startups; it locks startups into the CUDA ecosystem early, before they might consider alternatives; it creates recurring revenue streams that reduce dependence on cyclical hardware sales; it counters the hyperscaler threat by ensuring AI startups have an alternative to the big cloud providers that are also Nvidia's customers and potential competitors; and it aligns with the broader trend of revenue- and equity-sharing in AI, as seen with OpenAI's 5% stake offer to the U.S. government and the U.S. government's existing revenue-share positions in Intel and Nvidia's China sales [1][2][24].

---