Patrick Moorhead (@PatrickMoorhead) header
Patrick Moorhead (@PatrickMoorhead) profile picture

Founder, CEO, Chief Analyst @MoorInsStrat. Co-founder of @TheSixFiveMedia and @Signal_65. Ex-AMD CVP & Fellow, Compaq and AltaVista dot-com bomb participant.

Austin, TX
Joined May 2008
Photo by Patrick Moorhead (@PatrickMoorhead)Photo by Patrick Moorhead (@PatrickMoorhead)Photo by Patrick Moorhead (@PatrickMoorhead)Photo by Patrick Moorhead (@PatrickMoorhead)Photo by Patrick Moorhead (@PatrickMoorhead)Photo by Patrick Moorhead (@PatrickMoorhead)Photo by Patrick Moorhead (@PatrickMoorhead)Photo by Patrick Moorhead (@PatrickMoorhead)Photo by Patrick Moorhead (@PatrickMoorhead)Photo by Patrick Moorhead (@PatrickMoorhead)
The Week Ahead In Technology for November 16, 2026. $NVDA reports without Jensen Huang, and the numbers aren't my worry. It's a stock shock, not a business shock. Will there even be a guide? Monday I'm on @CNBC Squawk on the Street: NVIDIA and the Anthropic IPO. $NVDA: supplier commitments more than doubled to $279B, driven mostly by memory. Jensen's calls. Who makes the next one? @AnthropicAI: IPO expected to price near $1T. Only circular if nobody pays. Somebody is. @Microsoft Ignite: will we get more insights on how many customers pay for agents by usage, not seats? Everpure $P: bought components to secure supply; free cash flow went negative. Does it come back? @Lenovo: record 9.1% infrastructure margin. Does it hold as AI servers ship? Ian Buck, my top internal pick for NVIDIA CEO, gives its SC26 special address. Who's your pick?
It's The Week Ahead In Tech for November 9, 2026: @nvidia's board is investigating whether Jensen Huang was the target of a foreign intelligence effort, right before NVIDIA reports. The board sees no reason to believe NVIDIA secrets were compromised. Three House committee chairs reportedly want a closed-door briefing. Let the facts come in. Tonight I'm on BBC News talking $NVDA. Friday I'm on @CNBC Closing Bell Overtime previewing its earnings. $CSCO booked $9.3B of AI orders from the cloud giants last year. Is the gap to revenue supply or timing? $CRWV: its biggest customer went from 71% of revenue to 36 in a year. Do its borrowing costs reflect that yet?
6
1
1
9
8,521
Consider the source. The firm is short on $AMD.
Investors treated $AMD as the winner of Nvidia’s China crisis. The problem: AMD has a China problem of its own. Our investigation:
Readers added context they thought people might want to know
Hunterbrook reports AMD TurinPI firmware in post-Mar 2025 updates from Entity List-listed IEIT Systems (added 3/25/2025, denial policy) and linked firms. Report states it does not establish AMD violation or how firmware was obtained. Hunterbrook Capital is shorting $AMD. govinfo.gov/content/pkg/FR…
16
13
117
9,454
$LNVGY Q2 FY27: BEAT. The $640M loss comes from a $1.9B non-cash warrant charge, not the business. The numbers: -Revenue $30.2B, +48% YoY, vs $27.6B est: BEAT -Adjusted net income $1.36B, +166%, vs $1.27B est: BEAT -HKFRS net loss $640M vs $38M loss est: MISS From the call, beyond the documents: 1. Sunday I asked if the record 9.1% infrastructure margin holds as AI servers ship. It hit 9.6%. Plan: hold and grow it through Vera Rubin and Helios. 2. Over half of enterprise infrastructure now runs through the transactional model: the scale and profit I said it needed in August 2025. 3. PC market units fell ~20% last quarter. Double-digit declines, as guided in August, run 2 to 3 more quarters; gradual upturn from mid-2027. 4. Neoclouds: under 15% of infrastructure revenue last quarter, generally prepaid or title held until paid. The credit fear is overdone here. 5. Enterprise AI gear grew at a ~50% premium to traditional gear, much of which Lenovo says is bought for AI too. @Lenovo
2
1,528
Remember the doomer advertisements of yore from $ANT? IPO day it’s all positive AI vibes. Good to see but ironic.
2
10
2,337
Patrick Moorhead retweeted
Requested 2,000 $ANT shares, granted 143. Better something than nothing.
35
4
2
272
30,838
Requested 2,000 $ANT shares, granted 143. Better something than nothing.
35
4
2
272
30,838
$INTC Foundry has its first customer to officially announce EMIB, and it's $MSFT. @Microsoft will use @Intel_Foundry EMIB advanced packaging for Azure first-party silicon. To be clear, it's EMIB, not EMIB-T. There was some confusion. And the packaging piece is separate from the Xeon piece, even though both sit under one long-term agreement. What was announced: - EMIB packaging for Microsoft's own custom Azure silicon (Cobalt and or Maia) - Multi-year Xeon supply capacity, including next-gen Diamond Rapids - Xeon work in Azure on CXL memory, QuickAssist and TDX confidential computing - No terms, volumes or timing Some thoughts: -Packaging is the door into foundry. It's a smaller commitment than wafers and it goes straight at a real bottleneck. I've been saying the transistor isn't what limits AI anymore. Power, memory and packaging are. -Somebody had to go first. No one wants to be the first company to put its name on non TSMC, and now someone has. I'll be honest: I had Google and AWS as the likely first packaging announcements. Microsoft got there first. -This is the follow-through I was looking for. @LipBuTan1 has said he won't spend capex without a customer commitment, and on the October call he said a couple of packaging customers would be ready to announce in the next two quarters. Less than three weeks later, here's the first one. -The Xeon piece matters on its own. Agents need CPUs, and Intel says supply stays tight into 2028. Locking in multi-year capacity, Diamond Rapids included, is Microsoft buying supply insurance. -What's missing: terms, volumes and which chips exactly. Packaging gets Intel in the door. Wafers, 14A above all, are where foundry makes its money, and I still think that's 2029 or 2030. Congrats to the Intel Foundry folks.
Replying to @Microsoft
@Microsoft and Intel are expanding their decades-long collaboration, with plans to use Intel Foundry's EMIB advanced packaging technology to advance Azure infrastructure for cloud and AI. Learn more: ms.spr.ly/6019x1yiv #IntelFoundry #MSIgnite
27
35
2
270
15,392
Have we lost our freeking minds?!?
SITUATION DETECTED: The White House and Treasury are discussing a proposal that would require US companies to publicly disclose the use of Chinese open-weight AI models on their websites, aiming to discourage their use through public pressure, per Reuters.
89
57
5
1,148
148,312
Senator Tim Scott spewing nonsense on CNBC right now. My ears are bleeding it’s so dumb. Wrong on what China does with US tech, what’s law now, and even a complete lack of understanding that China cutting the US off from the basics would stall AI.
6
6
1
70
6,131
This @stratechery logic is flawed. Quarterly earning, the guide and ecosystem commitment number. were heavily influenced by Jensen. Nothing will change for a while due to the roadmap, supply chain commitments 18 months out.
“.. Nvidia’s earnings were boring, and also a statement as to why Jensen Huang probably isn’t coming back.” @stratechery $NVDA stratechery.com/2026/nvidia-…
2
2
26
3,283
$NVDA Q3 FY27 off the call and not in the release or CFO commentary. Top 5 things the call added: 1. FY28 growth outlook: ~75%, from 70%. ~$741B by my math vs $692B street: BEAT. 2. Saturday I gave Jensen's return a very small chance. The board doesn't intend to reinstate him and hired a search firm. 3. Kress: business as usual. Not for capital: suppliers first, customer financing selective, new commitments "rigorously" evaluated. 4. Q4 is the margin bottom: NVIDIA absorbed supply cost increases on booked orders, new pricing next year. FY28 held at 72% to 73%. 5. In May I wanted a clearer picture of the Rubin ramp: ~$20B in its first shipping quarter, hyperscalers in production in about a month. @nvidia
$NVDA Q3 FY27: BEAT and raise. Last week I called Jensen's exit a stock shock, not a business shock. NVIDIA raised its FY28 outlook. Exactly as I said. The numbers: -Revenue $115.8B, +103% YoY, vs $109.1B est: BEAT -Non-GAAP EPS $2.68 vs $2.48 est, $2.60 whisper: BEAT -Non-GAAP gross margin 74.9% vs 74% guide: BEAT. Q4 guide ~71.5% vs 71.7% est: MISS -Q4 guide ~$130B vs $125.3B est: BEAT Takeaways: 1. FY28 growth outlook: ~75%, from 70%. ~$741B by my math vs $692B street: BEAT. 2. The $130B guide is +12% QoQ, like the Q3 guide. Q4 has 14 weeks, so per week, ex-China, it's half: +6%. FY28's +75% needs ~18% a quarter. My math. 3. Monday I said we might hear Kress's sort order. We did: suppliers first, customer financing selective. Supply commitments: $390B. 4. Q4 is the gross margin bottom, as guided in August. FY28 stays below Q3 despite price increases: NVIDIA eats part of the memory bill. 5. In May I wanted a clearer picture of the Rubin ramp: ~$20B in its first shipping quarter, as guided. @nvidia
8
2
31
5,764
$NVDA Q3 FY27: BEAT and raise. Last week I called Jensen's exit a stock shock, not a business shock. NVIDIA raised its FY28 outlook. Exactly as I said. The numbers: -Revenue $115.8B, +103% YoY, vs $109.1B est: BEAT -Non-GAAP EPS $2.68 vs $2.48 est, $2.60 whisper: BEAT -Non-GAAP gross margin 74.9% vs 74% guide: BEAT. Q4 guide ~71.5% vs 71.7% est: MISS -Q4 guide ~$130B vs $125.3B est: BEAT Takeaways: 1. FY28 growth outlook: ~75%, from 70%. ~$741B by my math vs $692B street: BEAT. 2. The $130B guide is +12% QoQ, like the Q3 guide. Q4 has 14 weeks, so per week, ex-China, it's half: +6%. FY28's +75% needs ~18% a quarter. My math. 3. Monday I said we might hear Kress's sort order. We did: suppliers first, customer financing selective. Supply commitments: $390B. 4. Q4 is the gross margin bottom, as guided in August. FY28 stays below Q3 despite price increases: NVIDIA eats part of the memory bill. 5. In May I wanted a clearer picture of the Rubin ramp: ~$20B in its first shipping quarter, as guided. @nvidia
NVIDIA $NVDA Q3’27 EARNINGS HIGHLIGHTS 🔹 Revenue: $115.8B (Est. $109.3B) 🟢; +103% YoY 🔹 Adj. EPS: $2.68 (Est. $2.47) 🟢; +111% YoY 🔹 Data Center: $107.6B (Est. $101.9B) 🟢; +110% YoY 🔹 Adj Gross Margin: 74.9% (Est. 74%) 🟢; +140 bps YoY Q4 Guide: 🔹 Revenue: $130.0B +/- 3% (Est. $124.4B) 🟢 🔹 Adj Gross Margin: 71.0%-72.0% (Est. 71.7%) 🟡 🔹 Adj OpEx: ~$10.7B Segment Revenue: 🔹 Edge Computing: $8.2B (Est. $7.5B) 🟢; +42% YoY Capital Return: 🔹 Shareholder Returns: $28.1B in Q3 🔹 Buyback Authorization: Increased by $150B; no expiration 🔹 Remaining Authorization: ~$227.3B Key Updates: 🔹 FY27 GAAP & Non-GAAP Tax Rate: 16%-18%; excludes discrete items and material tax-environment changes Announcements: 🔹 Vera Rubin NVL72: Shipped; fastest product ramp in company history Comments: 🔸 “Vera Rubin is off to an explosive start, our roadmap is on track and our ambition has never been greater.”
2
2
1
9
6,973
At this point, the only fast track I see for a *quick* Jensen return is a mass top engineer walkout and work stoppage. Unlikely. What the board needs to consider is that they may get a slow drip of top talent out the door which accomplishes the same thing.
27
6
1
109
9,336
$MSFT Ignite came down to two words: cost and context. @Microsoft told customers Copilot does what ChatGPT and Claude do for less, while running their models and plugging its context into both. The model is now a component. That's the architectural disconnection between the model and every other stack layer I flagged in July. What was announced: - Copilot: Home, Office in Copilot and Managed Runtime GA. Code GA in December. Autopilot and Workspaces in Frontier, Today in private preview - Microsoft IQ: GA, no new SKU. Work IQ plugin in ChatGPT, Fabric IQ connectors for ChatGPT and Claude in preview - Models: seven, by Microsoft's count. MAI Code 2 in GitHub Copilot, Microsoft-Decision-1.1 GA, and MAI Licensed Weights so trusted customers can run a Microsoft model in their own Azure environment - Copilotforce with @salesforce: 38+ sales skills in Copilot, public preview - Agent 365: over 50M agents registered. Business Analytics and spend forecasting in Frontier - Silicon: @Arm-based Cobalt 200 VMs GA, up to 50% better performance than Cobalt 100, per Microsoft - Security: Microsoft Perception GA, core Integrated SOC in Defender GA My callouts: 1/ Cost was the keynote. @JasonTAndersen was in the room and heard that word more than any other. Microsoft's chart: $30 a month for an everyday Copilot user vs $81 for ChatGPT Enterprise and $205 for Claude Enterprise. Its advanced-user numbers matter more: $96 vs $149 vs $299. Moving to that user, with 20 agentic workflows a month, raises the Copilot bill $66 and ChatGPT's $68, by my math. Against ChatGPT, Copilot's edge is everyday use bundled into the seat, not cheaper agents. MAI Code 2 at 98% of GPT-6.1 Sol's coding performance for about 40% of the price is also Microsoft's number. The direction is still right: the cheapest correct answer wins. 2/ Context is the moat. Microsoft will take the context layer even where it doesn't own the chat window. Salesforce coming into Copilot says the same thing from the other side, and it keeps control of its data and permissions. The model labs want the work surface. They can now reach Microsoft's context. They still don't own Entra, Teams, Outlook and SharePoint. 3/ Agents need CPUs, as I said in October. Microsoft says Cobalt becomes a significant share of new Azure compute capacity in 2027, next to Maia 300 for inference. Own models on own silicon is the margin lever Microsoft needs to make agents cheaper, not just seats. Still missing: an updated paid Copilot seat count, how many customers pay for agents by usage, and credit burn for Autopilot and Code. A lot of this is still Frontier or preview. CFOs still need the number.
Highlights from my opening of the Ignite keynote this morning on how we’re building a frontier ecosystem to turn advances in AI into real-world impact.
17
2
2
53
7,303
Peak innovation.
10
40
3,766
German beer is, in fact, an innovation.
1
782
‼️
Over the past week, I’ve been overwhelmed by the love and support of friends, industry colleagues, and people I've never met. Thank you to everyone who has shown kindness to my family and me. It means the world to me. I came to the United States 53 years ago, and becoming an American citizen is one of the greatest gifts and proudest moments of my life. NVIDIA has been my life’s work. Building it into the company it is today alongside my colleagues has been a privilege and an honor. Nowhere but in America could a company like NVIDIA be possible. I love our country, and I love NVIDIA. I did not live up to the standards expected of everyone at NVIDIA, and I have made serious mistakes in my personal life. I created problems for the Board that I deeply regret. I own my mistakes completely, and I am committed to cooperating with the Board’s independent investigation. Since my departure, I’ve seen an outpouring of calls from NVIDIA’s employees, investors, business partners, and customers for me to return as CEO. Many continue to reach out to me directly as the company navigates complex technology, market, competitive, and geopolitical dynamics. NVIDIA is a once-in-a-generation company. If the Board of Directors decides that my return is in the best interest of our shareholders, our employees, the company, and America, I would be honored to answer these calls and return as CEO. Thank you again to everyone who has shared their good wishes. I am profoundly humbled by your support. To my family: I love you, and I will spend the rest of my life being worthy of yours. Jensen
11
1
120
12,817
Keep your eyes on this one.
Another big day. So excited to welcome the amazing @taalas_inc team to the @AMD family!
6
4
1
129
7,035
My home for the next 9 hours. This KLM flight is pretty good so far. Austin direct to Amsterdam.
23
107
5,990
Patrick Moorhead retweeted
Today's forecast revision from SemiAnalysis is pretty interesting. The TLDR is it calls for a 28% cut in AI accelerator capex in 2028 and then 41% and 40% cut the following two years respectively. The composition of these cuts was about 17% on shipments, and it has NVIDIA taking the brunt (66%) of the cut. Basically, demand in the past 10 days has been completely nuked? Remember, it was on election day that SA published this reiterating just how fast the build out is going. “We wrote at length about how SpaceX’s plans for 10GW in 2027 was feasible, paving the way for +300b+ ARR. We received a lot of pushback from folks calling it impossible. Now, just two months later, we’re already seeing 3.6GW under construction across multiple sites, both existing and new. SpaceX has even revised its estimate up to ~13GW in 2027. This would be the largest amount of datacenter capacity ever built by any one company in that timeframe, roughly equivalent to all of Google’s datacenter capacity from 1999-2026 in just 1.5 years. We think it’s possible and have already identified >9GW of capacity.” The forecast cut doesn't mention or reflect anything related to China exports or API use. But it does call for a significant slowdown in the growth of the frontier labs. This is also interesting in both the narrative and the timing given the Anthropic IPO this week. The report as interpreted indicates that HBM pricing will see double digit % price declines and ASPs on Feynman SKUs would take a 40% haircut from estimates with Feynman dropping from $110k b78k and estimates on Feynman Next dropping from $191k to $118k. A lot has to happen for demand to slow this much and for the semiconductor supply chain to normalize this much in a short period of time. While the report claims that its revision is based on 3 factors. 1. Lab monetization slow down. 2. Stricter access to to capital 3. Election-based data center delays. The revision is much harsher than a handful of GW being slowed by red tape. The report basically insinuates that the demand asymmetry is drying up in 2027. That in-house silicon projects are taking off at a dramatic pace, but only so much so that the memory wall will have been climbed. 66% of the revision down comes from NVIDIA. But again, why would its margins suffer so badly? In reality we have a lab IPO with $900 Billion in compute commits. We have uncertain demand in China, but no matter how much exports are restricted or tightened, it is hard to believe it will just stop seeking compute. And don't tell me NVIDIA demand will be replaced by Huawei Ascend. (It won't) It almost feels like SA is saying without saying that China is a MUCH BIGGER consumer of NVIDIA chips? And without it and its labs chewing up Inference on Anthropic and OpenAI and it using "NVIDIA" to train its models that the domestic demand will faulter substantially. If China really was utilizing a disproportionate amount of compute at peak prices this could cause a major downturn in margins for NVIDIA. Our forecasts (@ETRnews and @TheFuturumGroup) see much more modest scenarios with some newly minted downside risk. We agree that 27 is locked in and we see the same for most of 28. Projects into 29 and beyond have time for negotiation, but the compute demand for inference and agents will continue to climb dramatically. Some GW could slip and elongate the cycle. But seeing ASPs drop 40% on roughly the same volume suggests a structural weakening and a supply chain overcorrection or an underlying symptom that simply isn't reflected in lab demand, financing, or election outcomes. Perhaps China holds the key?
33
27
2
255
44,326
Thoughts on a Jensen comeback.
"After an exhaustive search, NVIDIA names its next CEO: Jensen Huang." I joked about it Thursday. What it takes and where investors fit with some Fable help: • Shareholders elect directors. Directors pick the CEO. Jensen's ~3.6% is big, not a majority. • Pressure can flip boards fast: nearly all @OpenAI staff threatened to walk and @sama was back in days. Here, a petition that's gaining steam. • Time is the ask from @altcap and @GavinSBaker who want a slow search with his return in play. • Tuesday cuts both ways. The quarter was Jensen's; the $NVDA call is Kress's. Steady may helps board. • The probes must go his way: @nvidia's review and Congress on Dec. 1 and 3. • A special meeting takes 15% held a year. @Vanguard_Group and @BlackRock alone top that. Index funds don't typically start fights. Maybe the slide and potential future slide motivates them. • No-cause director removal takes two-thirds of all shares. A board-backed 2025 fix got ~99% of votes cast and failed. • Binding route: nominate a 2027 board Feb. 24 to Mar. 26. Vote likely next summer. • What no letter fixes: a reported relationship with a direct report. Anything this misses? $NVDA
5
23
6,412