Every report the desk has filed, searchable by full text, topic, format, symbol and date range.
With Nvidia due to report and Jackson Hole on the calendar, rates and mega-cap tech risk are both back in focus for a market that has ridden the chip maker's results before.
Nvidia reports earnings Wednesday in what MarketWatch calls a test for a broad universe of companies tied to the AI theme, with bond yields also back in focus ahead of Jackson Hole.
Citi says the chipmaker's stock should rise after results, while another Wall Street voice warns it could fall even on a beat.
Citi expects Nvidia shares to trade higher after its upcoming earnings report, while a competing analysis argues the stock could drop even if the company beats estimates, because expectations have become too demanding.
Cloud giants are now outspending their own cloud revenue on data-center buildout just as Wall Street awaits Nvidia's next earnings report.
Major cloud companies are spending 102% of their cloud revenue on capital expenditures for AI infrastructure, a level flagged as unusual ahead of Nvidia's upcoming earnings.
Analysts are split on whether Nvidia's valuation is stretched or still cheap heading into its report.
Wall Street commentary ahead of Nvidia's upcoming earnings report is divided between arguments the stock is undervalued and warnings that the real test comes only after results land.
A broad chip-sector pullback hit Micron hardest, but analysts covering the stock say the drop looks overdone.
Micron shares fell 7% in a sector-wide semiconductor sell-off, prompting analysts to debate whether the drop is a buying opportunity.
The chipmaker is lining up one of the largest debt raises tied to the AI boom as custom silicon orders swell its backlog.
Broadcom is seeking up to $80 billion in debt financing to support an AI chip deal, one of the largest such raises tied to the current buildout of AI data centers.
Fractile's valuation leaps to $6.5bn on an Anthropic chip deal, SpaceX explores acquiring coding startup Cognition, Bloom Energy cuts fuel-cell installation time 40%, and Coatue's Laffont piles into semiconductors — evidence the AI trade is broadening beyond the incumbent GPU complex.
A cluster of transactions and product news illustrates how AI capital is migrating outward from established GPU vendors into custom inference silicon, distributed power generation, autonomous mobility and developer tooling. The breadth is a bullish structural signal but raises the bar on valuation discipline, with leveraged semiconductor vehicles flagged as poorly compensated risk.
Semiconductor selloff decouples from rates narrative while cybersecurity leaders CrowdStrike and Palo Alto Networks extend declines, signalling broad de-risking in 2026's most crowded growth trades
Intel and AMD each fell roughly 4% and NVIDIA finished flat as the semiconductor complex sold off even as Treasury long-end pressure eased, breaking the rates-driven correlation that has governed tech leadership. CrowdStrike sank 7% despite a Truist price target increase to $245, with Palo Alto Networks off 5%, evidence that positioning — not discount rates — is now the dominant driver.
BofA calls the pullback a 'compelling opportunity' while sell-side analysts argue the GPU replacement cycle is lengthening, as Nebius taps markets for $4.5bn in convertibles to fund AI infrastructure buildout.
A cluster of constructive AI-complex research landed simultaneously: Nvidia upgraded to Strong Buy on GPU cycle elongation and valuation, BofA flagging a 'deep discount,' Rosenblatt reiterating Buy on Cerebras, and Northland upgrading Ciena on strong supplier datapoints. Nebius's $4.5bn convertible offering underscores that neocloud capex financing remains wide open despite equity market skittishness.
Xiaomi's second-quarter print eases fears of smartphone-driven margin compression while Baidu and China Unicom slide, and reports of small Nvidia H200 shipments reaching China reopen the export-control question.
Chinese technology equities are dispersing violently on fundamentals rather than trading as a single macro block. Xiaomi rallied after handset pressure eased, while Baidu and China Unicom fell sharply. Separately, the Financial Times reports Nvidia H200 chips are arriving in China in small volumes, a potentially significant shift in the AI hardware access equation.
Lumentum, Fabrinet and SK hynix sit at the choke point of hyperscaler buildouts as optical transceiver and HBM shortages intensify — even as valuation math for the AI trade shifts
The binding constraint in AI datacenter construction is moving downstream from GPUs to optical interconnect and high-bandwidth memory, concentrating pricing power in a narrow set of suppliers. At the same time, sell-side analysis warns the arithmetic supporting AI valuations is deteriorating.
The sell-side pivot from thesis to price marks a maturing phase of the AI trade, coinciding with fresh evidence of concentrated semiconductor positioning among hedge funds and Google's willingness to pay cash for proprietary training data.
Goldman Sachs cut its rating on Nvidia, conceding AI exposure remains attractive while arguing valuation now warrants caution. The call lands alongside 13F disclosures showing Micron and SanDisk dominated one fund before a July collapse, and a $10 million Google purchase of Spirit Airlines data — signals that the AI capital cycle is broadening from compute into data and memory.
Beijing's flagship memory champion becomes the world's third-largest flash supplier while Google prepares to exit Pixel manufacturing in China by 2027, evidencing a two-way decoupling that is reshaping semiconductor pricing power.
YMTC's rise to third place in global NAND flash with a 14% share represents a durable structural shift in memory supply, threatening incumbent pricing discipline. Simultaneously, Google's reported plan to end Pixel production in China by 2027 illustrates that the decoupling is bidirectional, with Western OEMs relocating assembly as Chinese suppliers scale.
Off-balance-sheet obligations tied to compute leases, data centre financing and vendor structures are estimated at $3 trillion and expanding by $1.2 trillion per quarter, raising questions over reported free cash flow across the hyperscaler and neocloud complex.
Mainstream financial press has now adopted a thesis long circulated in independent research: the AI capital cycle is increasingly financed through structures that sit outside reported balance sheets. With obligations estimated near $3 trillion and compounding, the divergence between GAAP earnings quality at incumbents and cash-burning neoclouds is becoming the central risk debate of the cycle.
Memory pricing power lifts Micron into four-digit territory even as Nvidia's reduced guarantee on the OpenAI Ohio data center campus signals a more disciplined phase of AI infrastructure financing.
Micron's move above $1,000 per share underscores how the AI buildout has migrated from GPUs into the memory and interconnect layers of the stack. Simultaneously, Nvidia's decision to cut its guarantee tied to OpenAI's Ohio data center to $120 billion points to tighter vendor-financing discipline, a swing factor for the credit assumptions underpinning hyperscale capex. Blackstone's positioning as an AI infrastructure landlord and copper's squeeze-level pricing reinforce that the trade is now as much about physical inputs and financing structures as it is about silicon.
Bookrunner messaging points to an inaugural Australian dollar issue from the search giant, extending the megacap capital-raising cycle into secondary currency markets to fund AI infrastructure build-out.
Alphabet is sounding out an inaugural Australian dollar bond, according to a bookrunner message, marking the latest step in hyperscaler funding diversification as AI capital expenditure outstrips operating cash generation.
Tepper adds, Loeb exits entirely and Soros buys puts, exposing a widening institutional divide over whether hyperscaler capital spending is a durable earnings engine or a late-cycle peak.
Divergent positioning among high-profile investors in Nvidia — accumulation, full liquidation and outright put buying — highlights unresolved disagreement over the sustainability of AI infrastructure spending, even as Amazon, Alphabet and Microsoft commit close to $600 billion in capital expenditure.
Wall Street strategists shift narrative from 'spending risk' to 'spending-to-earnings conversion' as hyperscaler backlogs, memory pricing and semicap records validate the infrastructure buildout
Alphabet's raised 2026 capital expenditure guidance to $205 billion arrived alongside a $514 billion Google Cloud backlog, reframing the AI investment debate around revenue conversion rather than balance-sheet strain. Strategists now argue spending is directly feeding earnings, a thesis reinforced by SK Hynix's blowout week, record Applied Materials revenue and Nvidia's growing role as system-wide financier of the AI supply chain.
Analysts flag downside risk into Nvidia's print even as Fastly reports surging AI traffic and Micron and Sandisk emerge as split candidates — a widening dispersion inside the compute complex.
Positioning ahead of Nvidia's results is turning two-sided, while second-derivative AI plays in memory, edge delivery and analytics software attract fresh capital.
The chipmaker's largest-ever debt raise — more than triple its prior offering — lands despite $13bn in cash, underscoring the scale of capital intensity now required to compete in accelerated computing.
Advanced Micro Devices has tapped debt markets for $4.75bn, its biggest bond sale on record, even while holding roughly $13bn in cash and investments. The move fits a broader pattern in which AI infrastructure demand — from silicon to data center power — is forcing semiconductor and energy issuers alike to pre-fund multi-year capacity commitments.