#MarvellGoogleChipDeal

Marvell’s Google agreement matters less as a one-day share-price catalyst than as evidence that custom AI infrastructure is broadening beyond compute alone. The scope spans TPUs and related accelerators, storage controllers, networking and memory interfaces, giving Marvell several potential routes into Google’s buildout. The warrants, covering up to 58.97M shares at $206.58, are not a completed equity purchase. Their real analyti#BTCBreaks72K #FOMC9To3Split #PopMartEarningsWatch

Marvell’s Google agreement matters less as a one-day share-price catalyst than as evidence that custom AI infrastructure is broadening beyond compute alone. The scope spans TPUs and related accelerators, storage controllers, networking and memory interfaces, giving Marvell several potential routes into Google’s buildout. The warrants, covering up to 58.97M shares at $206.58, are not a completed equity purchase. Their real analytical value is alignment: if execution scales, expectations for custom-chip revenue may rise. The Aug 27 FY2027 Q2 report should show whether management is ready to translate that potential into guidance. Not advice, just analysis. #MarvellGoogleChipDeal

US pre-market action is looking strong—another big move could be coming tonight. $BTC and $ETH are both gaining momentum, with $67K BTC and $2K ETH looking like key levels this week. Meanwhile, $MRVL jumped 10% after news of a Google order. Storage names like $SNDK remain under pressure, adding more volatility to the market. I’m staying bullish. The signals still point north. 📈 #XiaomiQ2Earnings #SKHynix40TBuyback #SandiskValuationSplit

Real people sold $MRVL when it was under $200 😂 Now Google is buying

JUST IN: 🇺🇸 Google just secured the right to buy up to $12.2 billion of Marvell shares. The deal could generate up to $120 billion in qualifying Marvell revenue through 2033.

JPMorgan’s take on the $MRVL-Google deal (OW, PT $240): "We think it is important to clarify that we do not interpret the announcement as a core TPU accelerator win, as the custom silicon programs highlighted appear to be XPU attach/TPU-adjacent products (e.g., chips that sit next to the core TPU chip; MRVL referred to them as 'programs that attach to the TPU ecosystem'). Nonetheless, according to our math, the structure of the deal implies an average ~$19.2bn of potential Google ASIC attach revenues per year for MRVL through FY33 (240 tranches * $500m = $120bn of revenue over 6.25 years), which would put the team’s AI ASIC business well above its prior analyst day target of ~$11B in total AI ASIC revenues in CY28. We believe most of the revenue under this agreement is incremental to current market expectations, and after flowing the deal through our model, we can see a path toward ~$11.00 of earnings power by CY28, materially above current Street estimates of $9.52. Although the agreement does not represent a formal purchase commitment from Google, we believe it does showcase the potential scale of MRVL’s opportunity and could support positive upward revisions if/when purchases begin to ramp. We also think the announcement reflects Google expanding custom silicon development beyond the core TPU chip and into adjacent areas such as networking, storage, memory interfaces, and near-memory compute, where multiple partners can participate. Analyst: Harlan Sur

Marvell disclosed warrants giving Google the right to buy up to 58.97 million shares at $206.58 each, exercisable through August 2033. The warrants are tied to Google’s revenue target for custom-chip procurement and could represent about $12.2 billion at the exercise price if fully exercised. Google has not paid that amount or locked in all future orders. The arrangement links procurement scale with an equity incentive, potentially strengthening long-term cooperation and supply-chain certainty for custom chips. Why it matters: Major cloud providers are using capital tools alongside procurement contracts to secure capacity and make key supplier relationships harder to exit. Read the full article: https://t.co/TyPnROc5fn

$MRVL gives $GOOGL options to buy $12.2B of the company. "The Marvell-Google deal covers a broad range of technologies used ​with TPUs, including processors that run AI models, manage data storage ​and move information ⁠across networks." Which could translate into ~$120 billion ​in revenue through 2033. Marvell also has separate warrants with $AMZN after their Celestial acquisition (for purchases of photonic fabric). Remember the good times when Jensen said Marvell would be the next $1T+ company? Seems Nvidia knew Marvell was "networking" its way into all the hyperscalers with warrants.

Twenty-two years ago today, Google went public at $85 a share. I was working in enterprise technology in 2004. Search felt like a useful front door to the web—not the foundation of a company that would eventually span advertising, Android, YouTube, cloud, custom chips and autonomous vehicles. The IPO valued Google at just over $23 billion after $1.47 billion in 2003 revenue. Alphabet is now worth roughly $4.2 trillion, and its latest quarter produced nearly $120 billion. Search still pays for most of Google’s ambition. AI changes its role from directing people to information toward answering questions and completing tasks. That helps users but threatens publishers, merchants and advertisers built around clicks. Google must create new commercial formats without filling answers with sponsored clutter or starving the web of traffic. Cloud has the strongest case as an independent growth engine. Google can sell chips, computing, Gemini, data, security and agent tools even if Gemini never leads the consumer market. Enterprise customers, however, want choice across clouds and models—not another closed platform. YouTube is effectively Google’s media business—and arguably the largest media business in the world. It combines broadcasting, creators, search, streaming, subscriptions, connected TV and advertising. AI can improve discovery, dubbing and commerce. It can also flood YouTube with cheap synthetic content and raise questions about quality, creator compensation and trust. Android, Chrome and Workspace provide distribution across the phone, browser and working day. That supports adoption but invites scrutiny around defaults, bundling and market power. Waymo is Google’s most credible business outside software. Its challenge is operational: vehicles, fleets, insurance, regulation and public trust. It will not scale like an app. One lesson from three decades around enterprise technology: technical leadership and commercial adoption are different disciplines. Companies rarely standardize on whichever model wins this month’s benchmark. They choose what works with their data, identity systems, security policies and existing architecture. Google has the pieces, but its history of changing product direction and uneven reputation with enterprise buyers remain obstacles. Google’s advantage is not Gemini alone. It controls much of the chain from chips and data centers through models, applications, advertising and distribution. Those businesses can also work against one another. Better answers may reduce web traffic. Automation may reduce advertiser control. Synthetic media may alienate creators. Bundling may attract regulators. Google does not need another experiment. It needs to prove that Search, Cloud, YouTube, Workspace and Waymo become stronger together—and can carry some of the infrastructure bill. The question is no longer whether Google can build AI. It is what Google does next with it.