Olivia
Online
IntelBull
@IntelBull_
I invest in Tech Early in NVIDIA and Tesla -- bought it way before it was cool. Firm Intel bull—betting big as Lip‑Bu Tan steers Intel’s comeback 🚀
Joined November 2025
80 Following
675 Followers
1.1K Posts
Pinned Tweet
1/2 Everything you need to know about the $INTC turnaround.
My updated $200 price target thesis.
1. Earnings and Margin Explosion
Under Lip-Bu Tan, the turnaround is moving much faster than expected. The massive Q1 EPS crushed Wall Street consensus estimate, proving the fundamentals have shifted completely. The company delivered an impressive $13.58 billion in quarterly revenue, up 7.2% year-over-year, alongside reaching a massive 41.0% Non-GAAP gross margin. My old $110 price target was way too conservative, and Wall Street is finally waking up.
Evercore ISI just upgraded Intel to Outperform, adjusting their price target to $111 from $45.
Tigress also boosted their price target on Intel to $118 from $66, maintaining a Buy rating.
2. Complete CPU Dominance & Google Partnership
Intel still dominates the overall x86 CPU market with roughly 70-72% share. They maintain a massive 71% unit share in the critical Server and Data Center CPU market, which is essential for AI and agentic workloads. Client PC share remains a stronghold at 72-75%. Data Center and AI (DCAI) revenue just hit $5.1 billion, up 22% year-over-year. Beyond market share, Intel recently announced a massive multiyear collaboration with Google to advance the next generation of AI and cloud infrastructure. This partnership reinforces the critical role of Intel's CPUs and custom infrastructure processing units (IPUs) in scaling modern, heterogeneous AI systems across Google's global infrastructure.
3. Advanced Packaging, TeraFab, and Tesla
Intel's EMIB packaging is actually ahead of TSMC and represents the future of what they call a Systems Foundry approach. They are not just manufacturing wafers; they are dominating the entire packaging process. Wccftech just reported a massive foundry breakthrough, noting that Intel's EMIB has hit a 90% yield and that EMIB-T will scale to greater than 12x reticle in 2028. Current packaging revenue sits at $4-6B, mostly from internal EMIB and Foveros demand. Internal AI chiplet demand will grow 30-50% through 2026, and external foundry customers will drive a massive growth in 2027.
We are also seeing this scale in real-time with the TeraFab project with Tesla, which is set to utilize Intel's packaging capacity for Tesla's next-generation silicon. This TeraFab-level scaling secures their long-term path to becoming the dominant global foundry.
4. Big Tech is Moving to 18A & Expected Q2 Customer Announcements
Supply chain sources confirm that major players, including Google and Apple, have already started test chip verification for Intel's 18A process. Analyst Jeff Pu just backed this up, confirming that front-end 18A yields are doing well and consistent with management's comments. Even more massive: Pu expects Intel has already started 18A trial testing for an external customer's 2023 chip that was previously built on TSMC's N3B node. Moving through Q2, Wall Street is highly anticipating official announcements naming these major external foundry customers. Locking in these whale clients publicly in Q2 will serve as a massive catalyst that forces the market to completely re-rate Intel's foundry valuation.
5. The Nvidia Partnership
Jensen Huang directly confirmed Nvidia will use Intel's Foveros packaging for their upcoming x86 consumer CPU that utilizes Nvidia's GPU chiplets. Predictions point to Nvidia replacing Grace with a co-designed x86 CPU featuring 288 or more E-cores. They will need Intel's advanced packaging for Project Feynman, stacking SRAM and I/O dies on 18A/14A nodes. Furthermore, Intel's 14A 1.0 PDK will have a sizable advantage over TSMC's 2nm, giving Intel a one-year head start since TSMC's A14 is expected to be delayed past 2028.
JUST IN: $INTC Intel and Hitachi $HTCI.NE announce a strategic collaboration to accelerate AI transformation across manufacturing, energy, transportation & key industries.
Hitachi, a global industrial tech leader in OT/IT, manufacturing & infrastructure, brings deep expertise and manufacturing know-how.
Intel will be contributing with advanced computing, silicon photonics & purpose-built silicon.
The two will be focused on Foundry tools, quantum computing, energy optimization, custom chips, edge AI & factory automation.
Another vertical ecosystem win for Intel’s full-stack AI push. It validates the agentic / physical AI infrastructure story.
$INTC Sold my net week $140 calls and just DCAed all morning into 06/18/2026 $150 calls. Today's dip was a wonderful opportunity. Now at a profit. Expecting a run up soon!
Hon Hai (Foxconn) and Intel have announced a strategic collaboration to advance next-generation AI infrastructure, Edge AI, and Physical AI platforms.
✨ Read more: https://t.co/19Rpfyu3so
#Intel #AI #EdgeAI #PhysicalAI #AIInfrastructure #HonHaiTechnologyGroup
$INTC I’m not sure what it is, but there is a much bigger news coming in for Intel and very soon.
A rack-scale partnership.
Foxconn says it has entered into a strategic partnership with Intel.
I genuinely believe these could be the last opportunities to buy $INTC at current levels. When you look at Intel’s ongoing transformation, investments, and long-term strategy, the stock is significantly undervalued relative to what they have accomplished and how well they are executing.
Meanwhile, institutional investors seem to be quietly accumulating shares while sentiment remains mixed. That’s often when the biggest opportunities are created.
Don’t overlook Intel. When momentum returns, the next move higher will happen quickly, just as we saw last year and again a few months ago. By the time the market fully recognizes the value, these prices will already be gone.
Stop underestimating Intel which is the most overlooked chip company in the industry.
Google is diluting shares to raise $80 billion for AI infrastructure. A company with over $130 billion in annual net income faces compute capacity constraints severe enough to require equity issuance to meet enterprise demand.
Hyperscalers continue aggressive capacity expansion. Market pricing does not fully reflect Intel’s position.
At Computex 2026, Intel announced and detailed:
• Clearwater Forest (Xeon 6+) processors on the 18A process node, with up to 288 E-cores, addressing power density and efficiency needs in server racks.
• Rack-scale AI infrastructure solutions, including production-ready racks combining Intel Xeon processors with SambaNova SN-50 RDUs for high-performance, cost-efficient AI inference and agentic workloads.
$INTC Foundry Services (IFS) remains undervalued. TSMC cannot supply 100% of global advanced chip demand long-term due to capacity constraints.
Microsoft, Amazon, Google, and Ericsson are confirmed IFS customers for custom silicon! Again people in the industry are paying attention but not Wall Street. I repeat Microsoft, Amazon, Google, and Ericsson are confirmed IFS customers for custom silicon!!!
Google’s $80 billion raise indicates sustained high AI infrastructure spending. Intel offers immediate enterprise hardware (CPUs, rack solutions) plus U.S.-based foundry capacity for supply chain diversification.
Intel is extremely undervalued compare to the size of the opportunity ahead
This is huge for $INTC, but I am not surprised. As I said, people underestimated $INTC advanced packaging capabilities
So now $INTC secured wafer orders from Apple and Google and licensed 14A to Tesla/SpaceX. Who is the next?
$INTC CFO publicly stating that foundry progress is likely ahead of schedule by a little over a quarter. At BofA tech conf.
JUST IN: MediaTek says its next-gen program will only use $INTC Intel’s EMIB-T packaging, with tape-out by Q4 this year and mass production Q4’27.
This statement from Goldman Sachs Taiwan Day, aligns with reports from March this year of Mediatek and Intel partnerships to build $GOOGL TPUs.
Intel will be an absolute packaging powerhouse $INTC
Intel details long-awaited Crescent Island AI GPU at Computex, boasts up to 480 GB of LPDDR5X https://t.co/TLvIWfoOoC
Now $INTC at Computex 2026:
1/ The repositioning and the four-ecosystem thesis.
LBT did not open with a product. He opened with a thesis. AI is splitting compute into four ecosystems, PCs, the edge, foundational data centers, and emerging "intelligence centers" for digital agents. Each needs purpose-built CPUs, GPUs, and ASICs. The throughline is that agentic inference moves the data center from roughly one CPU per eight GPUs toward parity. Right processor for the right workload. I've pubicly espoused the benefits and caveats of heterogeneous compute since my AMD days.
Shift to inference is a shift I called in June 2025 on X as this happened 8 years ago with AI v1, so I read the pivot as correct, not convenient. Easy call. Not chasing NVIDIA on training concentrates Intel where the ratio moves in its favor and x86 compatibility still counts. The risk is narrative as the market still scores AI by GPU share.
2/ Xeon 6+ "Clearwater Forest" at rack scale. [Prior chip, previewed October 2025 and detailed at MWC in March; rack-scale framing new here.]
Xeon 6+, the Clearwater Forest E-core part, is 288 cores and 576 MB of L3 on Intel 18A. Intel did not unveil it here. It gave a first look at Intel Tech Tour in October 2025, detailed the full architecture at MWC in March, and put it into production about two weeks ago (Intel, Oct 2025).
What is new at Computex is the rack-scale framing, more than 36,000 cores in 32U and up to roughly 150,000 agents per rack, with ODM systems shipping today. Ericsson, a launch partner, put the gain at 60 percent better performance per watt against the prior-generation Sierra Forest on its own testing, and Intel showed a new Ethernet controller alongside.
This is a test of the call I went on record with in December 2025, that 18A would ship tens of millions of leading-edge parts and stand against any node in its class. A high-volume data center CPU on 18A, next to Panther Lake, is the external proof.
3/ Intel names Google and Ericsson as custom-silicon customers. [New customers; the ASIC business was stood up last October.]
The entry is not new, the customers are. Intel formally stood up its ASIC and chip-design business last October under a Central Engineering Group led by Srini Iyengar, the same leader who presented this segment on stage, and Tan was talking custom silicon at Intel Vision back in March 2025 (Moorhead, X, Mar 2025, Intel via SemiWiki, Oct 2025). What was missing was a customer.
In December I called the effort slideware and said I needed a sign of life from Intel custom ASIC (Moorhead, X, Dec 2025). This week delivered it. Intel is building another IPU for Google, a follow-on that is already designed and deploying, This is the way I read it. Ericsson picked Intel for next-generation telco silicon at "global scale", and there is vertical work at Echo Neurotechnologies, Greenstone Biosciences, Hitachi, and Siemens.
That is the sign of life, and another validation yet for Intel Foundry, attacking the franchise Marvell and Broadcom built from a design-plus-foundry position.
My colleague @MattKimball_MIS draws the right line: there is a demarcation between custom and merchant silicon, so Marvell and Broadcom are not suddenly direct competitors (Kimball, MI&S, Apr 2026). My caution stands: a Google IPU and an Ericsson telco part are adjacent infrastructure, not the hyperscaler training XPUs where the largest dollars sit. Credible start, not parity.
4/ Disaggregated inference: SambaNova, Foxconn, Vector Core Compute. [New customer proof; Intel's disaggregation thesis dates to last October.]
@SambaNovaAI's @RodrigoLiang demoed what Intel called the first heterogeneous disaggregated inference. One workload split across Intel Xeon 6 for orchestration, SambaNova RDUs for decode, and NVIDIA GPUs for prefill, measured by SambaNova and Artificial Analysis at two to three times GPUs alone.
Foxconn committed to joint rack-scale products on Xeon. Vista Equity's Robert Smith launched Vector Core Compute, a disaggregated-inference neocloud, with https://t.co/Esq3fulKeO as first customer and 50-plus US deployments planned. None of this is a new idea for Intel. It would have been great to see the dollar commitment.
Previous CTO @sk7037 laid out the thesis at OCP last October, calling for heterogeneous systems that match the right silicon to the right task on an open stack, with the Crescent Island inference GPU as Intel's piece. The SambaNova blueprint followed in April, when I called heterogeneous compute the way forward, cited in Data Center Knowledge, Apr 2026. The honest caution, which I gave Intel directly was that the marquee speed is mostly NVIDIA and SambaNova silicon, and Intel's own Crescent Island does not sample until the second half of this year, so no Intel accelerator was on stage to carry that load.
5/ Client and edge momentum, plus Perplexity on-device. [Momentum.]
Intel's 18A client lineup reached nearly 400 designs in a few months, with 300-plus Core Ultra Series 3 shipping. The Arc G3 handheld runs more than 40 percent faster than competition, or the same at half the power, available late this month. @perplexity_ai 's @AravSrinivas ran hybrid agentic inference, keeping a deal's confidential files on a local model on the Core Ultra Series 3 GPU. At the edge, Intel has 130-plus Series 3 designs and 4,000-plus partners, aiming at a physical-AI market it pegs at 25 trillion dollars by 2050. Client volume is the cash engine and the proof of 18A at scale, the same reason my December 18A call tied to high-volume client shipments. I was surprised to see @AlexKatouzian on stage this early but why not?
$INTC
![PatrickMoorhead's tweet photo. Now $INTC at Computex 2026:
1/ The repositioning and the four-ecosystem thesis.
LBT did not open with a product. He opened with a thesis. AI is splitting compute into four ecosystems, PCs, the edge, foundational data centers, and emerging "intelligence centers" for digital agents. Each needs purpose-built CPUs, GPUs, and ASICs. The throughline is that agentic inference moves the data center from roughly one CPU per eight GPUs toward parity. Right processor for the right workload. I've pubicly espoused the benefits and caveats of heterogeneous compute since my AMD days.
Shift to inference is a shift I called in June 2025 on X as this happened 8 years ago with AI v1, so I read the pivot as correct, not convenient. Easy call. Not chasing NVIDIA on training concentrates Intel where the ratio moves in its favor and x86 compatibility still counts. The risk is narrative as the market still scores AI by GPU share.
2/ Xeon 6+ "Clearwater Forest" at rack scale. [Prior chip, previewed October 2025 and detailed at MWC in March; rack-scale framing new here.]
Xeon 6+, the Clearwater Forest E-core part, is 288 cores and 576 MB of L3 on Intel 18A. Intel did not unveil it here. It gave a first look at Intel Tech Tour in October 2025, detailed the full architecture at MWC in March, and put it into production about two weeks ago (Intel, Oct 2025).
What is new at Computex is the rack-scale framing, more than 36,000 cores in 32U and up to roughly 150,000 agents per rack, with ODM systems shipping today. Ericsson, a launch partner, put the gain at 60 percent better performance per watt against the prior-generation Sierra Forest on its own testing, and Intel showed a new Ethernet controller alongside.
This is a test of the call I went on record with in December 2025, that 18A would ship tens of millions of leading-edge parts and stand against any node in its class. A high-volume data center CPU on 18A, next to Panther Lake, is the external proof.
3/ Intel names Google and Ericsson as custom-silicon customers. [New customers; the ASIC business was stood up last October.]
The entry is not new, the customers are. Intel formally stood up its ASIC and chip-design business last October under a Central Engineering Group led by Srini Iyengar, the same leader who presented this segment on stage, and Tan was talking custom silicon at Intel Vision back in March 2025 (Moorhead, X, Mar 2025, Intel via SemiWiki, Oct 2025). What was missing was a customer.
In December I called the effort slideware and said I needed a sign of life from Intel custom ASIC (Moorhead, X, Dec 2025). This week delivered it. Intel is building another IPU for Google, a follow-on that is already designed and deploying, This is the way I read it. Ericsson picked Intel for next-generation telco silicon at "global scale", and there is vertical work at Echo Neurotechnologies, Greenstone Biosciences, Hitachi, and Siemens.
That is the sign of life, and another validation yet for Intel Foundry, attacking the franchise Marvell and Broadcom built from a design-plus-foundry position.
My colleague @MattKimball_MIS draws the right line: there is a demarcation between custom and merchant silicon, so Marvell and Broadcom are not suddenly direct competitors (Kimball, MI&S, Apr 2026). My caution stands: a Google IPU and an Ericsson telco part are adjacent infrastructure, not the hyperscaler training XPUs where the largest dollars sit. Credible start, not parity.
4/ Disaggregated inference: SambaNova, Foxconn, Vector Core Compute. [New customer proof; Intel's disaggregation thesis dates to last October.]
@SambaNovaAI's @RodrigoLiang demoed what Intel called the first heterogeneous disaggregated inference. One workload split across Intel Xeon 6 for orchestration, SambaNova RDUs for decode, and NVIDIA GPUs for prefill, measured by SambaNova and Artificial Analysis at two to three times GPUs alone.
Foxconn committed to joint rack-scale products on Xeon. Vista Equity's Robert Smith launched Vector Core Compute, a disaggregated-inference neocloud, with https://t.co/Esq3fulKeO as first customer and 50-plus US deployments planned. None of this is a new idea for Intel. It would have been great to see the dollar commitment.
Previous CTO @sk7037 laid out the thesis at OCP last October, calling for heterogeneous systems that match the right silicon to the right task on an open stack, with the Crescent Island inference GPU as Intel's piece. The SambaNova blueprint followed in April, when I called heterogeneous compute the way forward, cited in Data Center Knowledge, Apr 2026. The honest caution, which I gave Intel directly was that the marquee speed is mostly NVIDIA and SambaNova silicon, and Intel's own Crescent Island does not sample until the second half of this year, so no Intel accelerator was on stage to carry that load.
5/ Client and edge momentum, plus Perplexity on-device. [Momentum.]
Intel's 18A client lineup reached nearly 400 designs in a few months, with 300-plus Core Ultra Series 3 shipping. The Arc G3 handheld runs more than 40 percent faster than competition, or the same at half the power, available late this month. @perplexity_ai 's @AravSrinivas ran hybrid agentic inference, keeping a deal's confidential files on a local model on the Core Ultra Series 3 GPU. At the edge, Intel has 130-plus Series 3 designs and 4,000-plus partners, aiming at a physical-AI market it pegs at 25 trillion dollars by 2050. Client volume is the cash engine and the proof of 18A at scale, the same reason my December 18A call tied to high-volume client shipments. I was surprised to see @AlexKatouzian on stage this early but why not?
$INTC](https://pbs.twimg.com/media/HJ0Q3ckaEAAnbta.jpg)
@imnotharsh Great summary!
Stop underestimating Intel which is the most overlooked chip company in the industry.
Google is diluting shares to raise $80 billion for AI infrastructure. A company with over $130 billion in annual net income faces compute capacity constraints severe enough to require equity issuance to meet enterprise demand.
Hyperscalers continue aggressive capacity expansion. Market pricing does not fully reflect Intel’s position.
At Computex 2026, Intel announced and detailed:
• Clearwater Forest (Xeon 6+) processors on the 18A process node, with up to 288 E-cores, addressing power density and efficiency needs in server racks.
• Rack-scale AI infrastructure solutions, including production-ready racks combining Intel Xeon processors with SambaNova SN-50 RDUs for high-performance, cost-efficient AI inference and agentic workloads.
$INTC Foundry Services (IFS) remains undervalued. TSMC cannot supply 100% of global advanced chip demand long-term due to capacity constraints.
Microsoft, Amazon, Google, and Ericsson are confirmed IFS customers for custom silicon! Again people in the industry are paying attention but not Wall Street. I repeat Microsoft, Amazon, Google, and Ericsson are confirmed IFS customers for custom silicon!!!
Google’s $80 billion raise indicates sustained high AI infrastructure spending. Intel offers immediate enterprise hardware (CPUs, rack solutions) plus U.S.-based foundry capacity for supply chain diversification.
Intel is extremely undervalued compare to the size of the opportunity ahead
Intel is the next trillion dollar company.
Good luck @LipBuTan1 tomorrow! Do us proud! The whole Intel community is behind you! 🤗
I’m in Taiwan and will try to attend Computex!
😆 LOL at this $INTC reaction to $NVDA PC chips that will be used in Surface Laptops for consumer use is absolutely hilarious.
Are we being serious here?
Thanks for the discount you absolute geniuses. 🙄
Most Popular Users
Elon Musk 
@elonmusk
240.2M followers
Barack Obama
@barackobama
119.3M followers
Donald J. Trump
@realdonaldtrump
111.6M followers
Cristiano Ronaldo
@cristiano
108.9M followers
Narendra Modi
@narendramodi
107M followers
Rihanna
@rihanna
97.3M followers
NASA
@nasa
92.1M followers
Justin Bieber
@justinbieber
90.6M followers
KATY PERRY
@katyperry
86.8M followers
Taylor Swift
@taylorswift13
80.6M followers
Lady Gaga
@ladygaga
72.2M followers
Kim Kardashian
@kimkardashian
69.4M followers
YouTube
@youtube
68.6M followers
Virat Kohli
@imvkohli
68.6M followers
Bill Gates
@billgates
63.4M followers
The Ellen Show
@theellenshow
62.5M followers
CNN
@cnn
61.9M followers
Neymar Jr
@neymarjr
61.1M followers
X
@x
60.9M followers
Selena Gomez
@selenagomez
59.9M followers