Advertisement|Remove ads.

Shares of Intel (INTC), SanDisk (DNSK) and Aehr Test Systems (AEHR) surged to fresh 52-week highs on Thursday as investors doubled down on AI infrastructure winners benefiting from rising memory prices and hyperscale demand momentum.
Over the past year, INTC, SNDK and AEHR have surged 187%, 2,228% and 686%, respectively, underscoring strong investor interest across AI infrastructure semiconductor stocks.
Investor sentiment toward all three stocks has also remained firmly positive. On Stocktwits, retail sentiment across INTC, SNDK and AEHR stayed within ‘bullish’ and ‘extremely bullish’ territories amid high message volumes.
INTC shares gained after announcing a multiyear partnership with Google Cloud centered on next-gen AI infrastructure powered by Xeon processors and infrastructure processing units (IPUs).
Google deploys Xeon processors across instances supporting training coordination, inference workloads and general-purpose computing, while the companies are expanding co-development of ASIC-based IPUs designed to offload networking, storage and security tasks from CPUs.
Cantor Fitzgerald raised its price target on Intel to $60 from $45 on Thursday, still implying a slight downside from current levels and maintained a 'Neutral' rating. The brokerage said demand across compute remains “as robust as ever,” and valuations today appear “frankly disconnected” from the underlying fundamentals following the semiconductor sector's pullback.
The firm added that compute companies could be among the early beneficiaries if investors shift back toward a broader risk-on environment.
Intel has also drawn investor attention through partnerships with SambaNova Systems on heterogeneous inference architectures and with SpaceX, xAI and Tesla on silicon fabrication tech tied to Project Terafab.
SanDisk shares hit a record high on Thursday after receiving multiple analyst price-target increases, driven by strong NAND pricing trends and improving demand visibility across AI infrastructure deployments.
Cantor Fitzgerald raised its price target on SanDisk to $1,000 from $800 and maintained an 'Overweight' rating, implying a 17% upside from the stock’s last close. The firm said the broader semiconductor setup remains “AI hot vs everything else not” and added that following recent de-grossing and a roughly 15% selloff in the Philadelphia Semiconductor Index (SOX) through the end of the March quarter, the shift back toward risk-on positioning could come “NOW or SOON.”
Bernstein also lifted its target to $1,250 from $1,000, representing a 47% upside from current levels, while reiterating an 'Outperform' rating. The brokerage said NAND pricing continues to “surprise to the upside” with further acceleration expected.
The announcement of a two-week ceasefire framework to reopen the Strait of Hormuz has also been taken as a signal for investors to rotate back into AI infrastructure suppliers, particularly memory stocks.
UBS estimates that double data rate (DDR) memory prices rose 95% sequentially in the first quarter, while NAND flash prices increased 80%, and expects DDR prices to rise another 37% in the second quarter, with NAND pricing increasing 40%, according to a Barron’s report.
Aehr Test Systems is on track for its best weekly session since 2021, as investors focus on improving bookings momentum and on exposure to semiconductor reliability testing for AI and silicon photonics.
The company reported third-quarter bookings of $37.2 million and expects a significant near-term production follow-on order from a lead hyperscale customer. It also said bookings in the second half of this year are likely to land at the high end of its previously guided range of $60 million to $80 million.
Lake Street Capital Markets raised its price target to $56 from $50 and said the key development this quarter was the bookings inflection. William Blair also noted that Aehr has secured major AI and silicon photonics customers and expects additional announcements that could continue to boost shares.
Competition across the AI semiconductor ecosystem is increasing as major tech companies expand their internal chip development efforts to reduce reliance on third-party suppliers.
Anthropic is reportedly exploring the possibility of designing its own AI processors as demand for its Claude models accelerates and compute requirements scale rapidly. The move mirrors broader industry efforts by large model developers to secure dedicated infrastructure capacity.
At the same time, Amazon CEO Andy Jassy told CNBC this week that all Anthropic models currently in production were trained on AWS Trainium chips, underscoring its growing role in large-scale AI workloads. Jassy added that Trainium 2 capacity has already sold out, Trainium 3 is nearly fully subscribed and part of Trainium 4 capacity is already reserved.
Amazon’s chip unit is now operating at a $20 billion annual revenue run rate, and the company has signaled it could eventually sell processors externally as demand for AI continues to strain supply.
For updates and corrections, email newsroom[at]stocktwits[dot]com.