Intel Surges: CPU Narrative Heats Up! Is It Still Time to Buy Semiconductors?
This week, both $Intel(INTC)$ and $Advanced Micro Devices(AMD)$ surged, stealing the spotlight from the previously hot memory sector. Although SK Hynix delivered very strong earnings, SanDisk in the memory sector declined yesterday.
Why has CPU replaced memory as the new favorite?
Agentic AI workloads (task scheduling, state management, I/O control) are overwhelming GPUs alone — CPUs are now critical infrastructure again, not just supporting hardware.
Over the past two years, the market has been used to a single chain: capex up → GPU orders up → HBM up → advanced packaging up.
In this chain, CPUs were just a supporting role — “one extra chip bundled in the server,” and were given low valuation weight.
Morgan Stanley’s March AI Agent report pointed out:
pure inference and agent-based inference are physically different, and putting them into the same BOM model ignores the CPU as a core driver.
GPU is the BRAIN, responsible for token generation; CPU is the ORCHESTRATION layer, responsible for task coordination; Memory is KNOWLEDGE, responsible for data retrieval
Investors, take note! What are the four major investment opp ...
Every tool call in an agent workflow must first be processed by the CPU:
parsing JSON outputs, reconstructing context and the next prompt, then sending it back to the GPU for the next round of inference.
It may look like simple string processing, but under long context windows + multi-step reasoning, each round processes tens of KB to several MB of intermediate state.
$Intel(INTC)$: Best Quarter, Eyeing a 26-Year High
Intel delivered far stronger-than-expected sales guidance. A year ago, the market was asking if Intel could survive. Now, it’s asking whether Intel has enough capacity.
The company stated that June-quarter revenue is expected at $13.8B–$14.8B, versus analyst expectations of $13B.
-
Data center revenue surged
-
Cloud customers are aggressively buying Intel’s latest processors
-
Foundry business: revenue per customer guidance upgraded from “hundreds of millions” to “billions”
-
One-time restructuring charges still exist, but core business is clearly turning around
$CSOP SK Hynix Daily (2x) Leveraged Product(07709)$: Memory Scarcity Is No Longer Just a Narrative
Goldman Sachs’ takeaway after 1Q26 results:
-
Revenue +198% YoY, operating profit +405% YoY
-
DRAM ASP forecast raised to +182% YoY (from +144%)
-
HBM demand is fully booked for the next 3 years, supply cannot meet LTA commitments
-
ROE projected at 60–70%, a record high for SK Hynix
-
Target price raised to ₩1,800,000
JPMorgan’s data showing DRAM contract prices up 96% QoQ confirms this is not a short-term spike — the memory cycle has shifted into a new gear.
After the surge, are semiconductors overvalued or not?
CPU vs Memory — Which AI Hardware Leg Has More Upside?
-
Do you think Intel can reach a new all-time high?
-
Is a $100 price target for Intel reasonable?
-
Will CPUs replace memory as the next hot sector?
-
Leave your comments to win at least 5 tiger coins~
Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.
I don’t see CPUs replacing memory—they’re becoming equally important. HBM demand is still tight, but CPUs were previously underappreciated in the AI stack, so this looks more like a catch-up trade than a full rotation.
On valuation, I’d stay selective. $Intel(INTC)$ can keep rerating if execution holds, but a move to $100 likely needs sustained data center strength. I stay constructive overall, but prefer buying pullbacks rather than chasing.
@Tiger_comments @TigerStars @TigerClub
For a while there Intel looks like it was losing its rhythm but now the critics are silenced with an astounding earnings report.
Intel reached a record high of USD 85.22 on April 24 2026. This rally has eclipsed the previous all time high set on August 31 2000.
The surge is driven by a fundamental shift in Intel's trajectory due to unprecedented AI server demand for server CPUs to support agentic AI, fueling a CPU renaissance.
Successful high volume manufacturing on the 18A node has proven Intel can execute its technical road map.
Q1 2026 adjusted EPS of USD 0.29 crushed the USD 0.02 analyst estimate, showing significant margin improvement.
Can Intel go higher?
KeyBanc recently raised its price target to USD 110 & HSBC increased its target to USD 95.
What an amazing turnaround for Intel! It is time to buy this stock.
@Tiger_comments @TigerStars @Tiger_SG
The company stated that June-quarter revenue is expected at $13.8B–$14.8B, versus analyst expectations of $13B.
A "$100" price target is an optimistic but achievable goal within a two-year horizon. Some analysts have already moved their targets to this level, betting on a "server CPU renaissance" and the expansion of AI PCs. However, this valuation requires Intel to maintain flawless execution in its data center segment and avoid any further delays in its product roadmap.
Intel recently reached a new all-time high of $83.62 on April 24, 2026, following a massive 24% single-day surge. This breakout surpassed its previous dot-com era peak of $75.81 from August 2000.
A $100 target is considered reasonable by several major firms following Intel's stronger-than-expected Q1 2026 earnings.
Bullish Consensus: Multiple analysts, including those from HSBC and Roth/MKM, have set targets at or above $100, with high estimates reaching $110.
Risks: Despite the record high, the stock's forward P/E is elevated, and many analysts still maintain a "Hold" rating, citing potential overvaluation and fierce competition from Nvidia and AMD.
CPUs are re-emerging as a critical AI component, though they aren't necessarily "replacing" memory; instead, they are shifting from being "left behind" to becoming a bottleneck.
Intel's path forward largely depends on its 18A process node yields and its ability to secure additional large-scale foundry customers.
On CPU vs Memory, I would not call it a replacement cycle. CPUs are becoming critical as the “brains” for AI orchestration and inference, while memory, especially HBM/DRAM, remains the bandwidth backbone. Both can run, but leadership may rotate.
My pick:
• Near term momentum: CPU
• Higher upside torque: Memory
• Safer long-term compounder: CPU
Intel at a new ATH? Possible.
Intel above US$100? Possible, but execution must be near flawless.
2. Yes above $40 is possible
3. No cpu silicon is a depreciating asset
4. Thankyou
CPUs will not "replace" memory, but they are reclaiming the spotlight as the next critical bottleneck. As AI moves from the training phase to the "inference" phase, the demand for high-performance CPUs is scaling rapidly. While memory remains essential for data throughput, CPUs are becoming the "hot sector" for investors looking for the next phase of AI infrastructure growth beyond the initial GPU and HBM craze.
Intel (INTC) has the potential to reach a new all-time high, provided it successfully executes its "18A" process node and regains manufacturing leadership. While the stock has seen a massive "211%" surge recently, its ability to surpass previous peaks depends entirely on reclaiming market share from AMD and proving that its foundry business can attract major external "whale" customers.
但这不代表内存不行。像SK海力士这种，基本面依然非常硬，HBM供不应求也不是故事。闪迪走弱，更像个股或短期情绪，而不是行业见顶。
Over the past two years, the market has been used to a single chain: capex up → GPU orders up → HBM up → advanced packaging up.
In this chain, CPUs were just a supporting role — “one extra chip bundled in the server,” and were given low valuation weight.
Morgan Stanley’s March AI Agent report pointed out:
pure inference and agent-based inference are physically different, and putting them into the same BOM model ignores the CPU as a core driver.