NVIDIA (NVDA) on Solana

Nvidia is scheduled to report fiscal Q2 earnings on August 26 after market close, with the consensus EPS forecast at $2.01 — more than double the year-ago quarter. Barchart options data on August 28-expiring contracts shows a put-to-call ratio of 0.69x and an upper strike above $226, implying options traders are positioned for a potential 5% or greater rally following the report. The stock was trading near $214.72, up roughly 15% year-to-date, with RSI in the low 50s and room to run before reaching overbought territory.

Analyst sentiment heading into the print is broadly constructive. The consensus rating sits at "Strong Buy" with a mean price target near $304, representing more than 40% upside from current levels. BMO Capital Markets has Nvidia as a "Top Pick" with a $340 target. The company trades at approximately 18x forward earnings — a multiple analysts consider reasonable given the $80 billion buyback program announced in May and the ramp of next-generation Vera Rubin NVL72 systems in the second half of 2026.

Contract manufacturers building server systems around Nvidia's Vera Rubin and Grace Blackwell AI chips have begun notifying major data center operators—including Microsoft, Google, and Oracle—that prices will rise more than 15% on systems shipped in early 2027, Bloomberg reported. The exact increase varies by chip generation and memory configuration. The hikes originate with soaring DRAM costs: Samsung, SK Hynix, and Micron, which collectively control most of the world's memory production, face AI infrastructure demand that is outrunning their supply capacity, giving them unusual leverage over the entire server supply chain.

Even Nvidia, which operates at roughly 75% gross margins, is passing these memory cost increases downstream rather than absorbing them. Hyperscalers such as Amazon, Microsoft, Google, and Meta are ultimately bearing the higher costs through the contract manufacturers that supply their data centers. The episode illustrates a structural tension in the AI buildout: while Nvidia retains a near-monopoly on high-performance AI accelerators and commands premium pricing from customers, its own component economics are increasingly dictated by memory suppliers whose bargaining power has grown sharply alongside AI infrastructure spending.

Elon Musk announced during SpaceX's August 4 earnings call that the company would build "exclusively on Nvidia" going forward, citing the Vera Rubin architecture as "the best architecture." The public commitment drove NVIDIA shares up 3.4% on no new financial results from the chipmaker itself — an illustration of how heavily the market weights visible customer loyalty from high-profile AI infrastructure buyers.

The announcement landed the same day AMD reported $11.5 billion in record quarterly revenue, underscoring the competitive stakes. Institutional preference already skews toward NVIDIA: 275 hedge funds held NVIDIA positions in Q1 2026 versus 134 for AMD. SpaceX's exclusive pledge reinforces that advantage by signaling that leading AI operators view NVIDIA's next-generation Vera Rubin hardware as the definitive platform for scaling compute-intensive workloads.

Nvidia on August 20 denied a report from The Information claiming the company planned to ship a China-customized LPU AI chip in small batches by year-end. The denial adds uncertainty just days before Nvidia's fiscal Q2 earnings release on August 26, where the company has guided for approximately $91.0 billion in revenue — roughly 11% sequential growth — with gross margins near 75%. Current guidance explicitly assumes zero revenue from Chinese data center compute sales, a posture CEO Jensen Huang reinforced in May when he acknowledged Nvidia had "largely conceded the country's AI chip market to Huawei."

The timing is notable: Nvidia received licenses in May to ship H200 chips to Alibaba, Tencent, and ByteDance, leaving open questions about future China revenue. SpaceX has separately committed $150–250 billion in Nvidia chip spending to reach 10 gigawatts of compute by end of 2027, providing a clear demand floor outside China. With hedge fund ownership rising to 275 funds and short interest minimal at 1.26% of float, the market enters earnings week watching whether management updates its zero-China-revenue assumption or signals any path back into that market.

Jensen Huang raised Nvidia's cumulative revenue target for its next-generation AI chip platform from at least $500 billion to $1 trillion through 2027, effectively doubling the company's own prior guidance. The revision, announced at Nvidia's March GTC conference, is driven by what Huang calls the "inference inflection" — a structural shift in AI workloads that puts inference demand alongside model training, expanding the total addressable market for Blackwell and the forthcoming Vera Rubin architectures. On August 10, Huang broadened the growth thesis further by announcing memoranda of understanding with Goldman Sachs, KKR, and BlackRock aimed at attracting more than $500 billion in external capital to finance AI compute infrastructure, framing Nvidia chips as "a new class of productive, investable infrastructure."

Current quarterly revenue around $91 billion puts Nvidia on a trajectory that makes $400 billion in annual revenue plausible if 12% sequential growth continues, though reaching $500 billion in the current fiscal year remains unlikely. Analysts covering the stock argue that the market is not pricing in an unrealistic premium given that multiple expansion has tracked earnings growth closely, and that shares should continue moving higher in line with sales and earnings as the inference build-out accelerates.

CEO Jensen Huang moved to defuse mounting "circular financing" accusations this week, arguing that Nvidia's $500 billion memorandums of understanding with six financial firms — Apollo, BlackRock, Blackstone, Brookfield Asset Management, Goldman Sachs, and KKR — represent genuine outside capital rather than the vendor-financing structures critics have flagged. "The demand is real," Huang said, adding that "the capital is not Nvidia revenue." The pushback targets an older deal pattern in which Nvidia takes equity stakes or provides debt guarantees to customers such as OpenAI and CoreWeave, who then use the proceeds to buy Nvidia chips — a loop that both the IMF and the Bank for International Settlements have identified as a systemic downside risk within the broader AI investment cycle. Bloomberg has reported $540 billion in Nvidia-linked financing arrangements struck in 2026 alone.

Wall Street analysts are divided on whether the new MOU-based platform changes anything fundamental. Bank of America's Vivek Arya called it "a pivot away from vendor-financing that drew circularity fire," and Morgan Stanley said the structure should "alleviate circularity concerns." Skeptics counter that directing third-party capital toward Nvidia's own customers is still a variant of the same loop. Credit markets have tracked the uncertainty closely: Nvidia's 5-year credit default swap spread surged to a record 82 basis points in late July, up from roughly 40 basis points at the start of the month, its largest single-day intraday gain since CDS trading began in November 2025. NVDA shares fell nearly 5% to $196.51 during the stretch, erasing roughly $250 billion in market value. Investors are expected to press for fuller disclosure at Nvidia's Q2 FY2027 earnings call on August 26.

Nvidia has signed memorandums of understanding with six major Wall Street firms — Apollo Global Management, BlackRock, Blackstone, Brookfield Asset Management, Goldman Sachs, and KKR — to establish financing platforms that let customers fund AI data center construction and chip acquisition using institutional capital rather than their own balance sheets. The arrangements collectively represent more than $500 billion in third-party capital, drawing on institutional credit, insurance funds, and private capital to underwrite GPUs and data center infrastructure for hyperscalers, frontier AI labs, and enterprises.

Speaking to CNBC, Nvidia CEO Jensen Huang framed the development as a structural shift in how capital markets categorize compute. "This is really the first time that technology chips have become an investable asset class," Huang said, describing Nvidia's GPUs as "revenue-generating assets" that are "productive, long-lived, fungible, and flexible." Because Nvidia hardware is broadly adopted and transferable across customers, lenders can treat compute as a reliably underwritable long-duration asset — applying logic similar to traditional equipment or infrastructure financing to GPU deployments.

Nvidia is committing $5 billion to take equity and project finance positions in large-scale AI compute infrastructure, splitting the total between a $3 billion stake in energy-focused firm Lancium and a $2 billion commitment to Firmus' Project Southgate. The move marks a deliberate expansion beyond hardware sales: rather than supplying accelerators to third-party operators, Nvidia is embedding itself directly into the ownership and economics of AI data center campuses, taking shared returns from equity, leases, and project finance alongside its existing hardware and software revenue. Specific infrastructure tied to these commitments includes Lancium's 200 MW Abilene site, Beacon Point Texas leases targeting 1 GW of planned capacity, a Volta Infra AI factory in Norway, long-haul fiber builds with Zayo, an SK Group 2 GW project, and Blackstone AI infrastructure partnerships.

The strategy embeds Nvidia's Vera Rubin accelerated computing stack into long-term AI factory commitments rather than selling into a fragmented market of buyers. Analysts note the trade-off: Nvidia gains deeper recurring exposure to AI compute demand but takes on capital-intensive build risk, power availability constraints, and tighter concentration among a smaller group of large AI buyers and operators. The infrastructure pivot follows Nvidia's existing dominance in data center GPU supply, where it holds over 95% market share, and signals the company is positioning for a broader role as AI compute transitions from early buildout toward industrialized, utility-scale operation.

ARK Invest purchased approximately $15 million in Nvidia shares across five ETFs on July 28, 2026 — with $8.1 million flowing through the ARK Innovation ETF — one day after Meta reported Q2 earnings and raised its 2026 AI infrastructure budget to $130–145 billion. ARK followed up the next day with roughly $14.7 million in Taiwan Semiconductor shares across four funds, treating TSMC as the downstream beneficiary given Nvidia outsources its chip manufacturing there. The purchases signal conviction that hyperscaler capital expenditure will translate directly into sustained GPU demand rather than tapering after an initial buildout phase.

Both Nvidia and TSMC were trading near 25x forward earnings at the time of the purchases, a valuation the piece describes as well below levels seen during earlier AI cycle peaks. The framing positions the buys as a rotation back into semiconductor infrastructure from the hyperscalers themselves, with Meta's willingness to absorb an earnings miss while raising capex guidance read as confirmation that AI infrastructure spending remains a multi-year commitment rather than a near-term flush.