AI Stocks: Fresh Bets From Meta to Sandisk

Meta’s multibillion-dollar push into AI infrastructure

Meta Platforms has paired strong recent share gains with a sharply higher spending outlook as it pursues what it calls personal superintelligence. The company expects capital expenditures to reach between $115 billion and $135 billion in 2026, up from $72 billion forecast for 2025. CFO Susan Li said the increase is driven by investment to support Meta Superintelligence Labs and the core business. Meta generated $116 billion in operating cash flow in 2025, underscoring how large the capex commitment is relative to its cash generation.

Management plans to direct the 11‑figure capex largely toward expanding AI infrastructure, including data centers and computing capacity, in order to power its services and reduce dependence on third parties such as Nvidia and cloud providers. CEO Mark Zuckerberg has said he expects increased productivity from AI to make advertising a meaningfully larger share of global GDP over coming years, contingent on Meta’s ability to keep improving AI tools that boost ad creativity, efficiency and return on ad spend.

Nvidia at the center of AI hardware and investor debate

Nvidia remains a focal point of the AI trade, with several developments highlighting both optimism and recent volatility. Morgan Stanley analyst Joseph Moore described the stock’s recent underperformance as out of step with strong near‑term fundamentals, citing very strong checks and growing optimism around earnings. Moore said he increasingly hears references to more than $9 in earnings power this year versus a $7.75 consensus, and argued that worries about competition from custom ASICs and AMD are overstated.

Nvidia plans to launch its Rubin GPU platform, the successor to its Blackwell chips, with the Vera Rubin architecture aimed at sharply reducing inference costs and GPU requirements for training large AI models. CFO Colette Kress has said the company already has visibility to around half a trillion dollars in Blackwell and Rubin revenue through the end of 2026, and has projected $3 trillion to $4 trillion in annual AI infrastructure spending by the end of the decade. Separately, Summa Corp. increased its Nvidia stake in the third quarter, with the stock now comprising about 50.4% of its portfolio, while multiple analysts maintain buy ratings and have raised target prices.

Other commentators see Nvidia as a leading candidate to be among the best‑performing AI stocks in 2026, pointing to rising agentic AI adoption, continued strong demand for Blackwell, and Taiwan Semiconductor Manufacturing’s plan to ramp advanced packaging capacity to 130,000 wafers per month by late 2026, of which Wedbush estimates Nvidia has secured roughly 60%.

AMD, Micron and Broadcom ride structural AI demand

Advanced Micro Devices is highlighted across several reports as both a high‑growth AI beneficiary and, based on PEG ratios, one of the cheaper AI stocks. Despite a forward price‑to‑earnings multiple near 40 and trailing earnings multiples above 100, analysts’ growth projections yield a PEG ratio around 0.5. AMD expects its AI data center revenue to grow by more than 80% annually over the next three to five years as it gains server CPU share from Intel and ramps its Instinct MI350 GPU line, which it calls the fastest‑ramping product in its history.

Beyond data centers, AMD reports strong momentum in PCs, with client segment revenue up 46% year over year in the third quarter and more than 120 Ryzen AI‑powered PC designs shipping in 2026. The company is also targeting edge computing, including supplying its Versal 2 adaptive SoC for Blue Origin’s Mark 2 lunar lander. Analysts expect AMD revenue to reach $62 billion by 2027 with earnings growing at a 45% annual rate.

Micron Technology is another preferred AI play, benefiting from tight supply in high‑bandwidth memory used in AI systems. Its PEG ratio is just under 0.7 and shares trade at about 12.3 times forward earnings. Micron has already contracted its entire 2026 HBM supply, and CEO Sanjay Mehrotra has said strong demand and supply constraints are creating tight conditions expected to persist beyond 2026. The company recently accelerated its forecast for the HBM market to reach $100 billion to 2028, implying roughly 40% annual growth.

Broadcom is cited as well positioned as hyperscalers develop custom AI chips, with an estimated 60% share in AI server ASICs and a $73 billion AI‑related backlog. Analysts suggest that networking upgrades for million‑GPU clusters should support demand for Broadcom’s Tomahawk 6 switches, potentially translating into strong revenue and earnings growth in 2026.

Sandisk and the AI-driven memory squeeze

Sandisk has emerged as one of the most extreme AI‑linked winners in the semiconductor space. Spun off from Western Digital in early 2025, the stock has returned about 1,500% since then, rising more than sixfold last year and more than doubling so far in 2026. The surge has been fueled by an unprecedented supply shortage in memory and storage tied to AI infrastructure demand, with memory prices reported to have risen 50% in late 2025 and projected to climb another 40% to 50% by the end of the first quarter of 2026.

Sandisk designs and manufactures NAND flash‑based storage for data centers and edge devices, operating a cost‑sharing joint venture with Kioxia that covers process technology and memory design. Its vertical integration, from wafers through to finished SSDs, is cited as a competitive advantage. The company is the fifth‑largest player in NAND flash but gained about 2 percentage points of market share in the 12 months to September 2025, while larger rivals Samsung and SK Hynix lost share. Several hyperscalers are currently testing Sandisk’s enterprise SSDs.

In its fiscal second quarter of 2026, Sandisk reported revenue up 61% year over year to $3 billion, driven by data center demand, and a 404% jump in non‑GAAP earnings to $6.20 per diluted share. Guidance for the third quarter called for revenue of $4.6 billion and non‑GAAP net income of $13.00 per diluted share at the midpoint, implying earnings more than doubling sequentially. Analysts have rapidly raised price targets, with a current median of $690 versus a pre‑earnings median of $400.

Analyst moves and lesser-known AI beneficiaries

Beyond headline chipmakers, analysts and investors are flagging additional AI‑linked names. Trimble, long known for geospatial positioning tools, now derives most of its business from software that supports autonomous vehicles, defense and robotics. The stock is described as trading at about a 10% discount after a recent pullback. High‑profile investors including Cathie Wood, via the ARK Space & Defense Innovation ETF, and billionaire Israel Englander, through Millennium Management, have recently bought shares.

Meanwhile, the AI supply chain is affecting Apple and its memory partners. Lynx Equity Strategy warned that Apple could face a sharper profitability squeeze than the market anticipates, citing channel checks that point to rising memory costs and an abrupt spike in NAND flash pricing after talks with long‑time supplier Kioxia reportedly broke down. According to Lynx, Kioxia may now be shipping less than Apple’s forecast demand, forcing the iPhone maker to seek alternative supply at higher prices just as AI‑related demand keeps memory markets tight.