AI Trading Ideas: See the AI Changes in Advance and Be the Winner in the Market
Hello everyone! Today i want to share some trading ideas with you!
1
$Alphabet(GOOGL)$ just introduced Gemini 3.7 Flash with introductory pricing of $0.75 per 1M input tokens and $3.75 per 1M output tokens.
Gemini 3.7 Flash pairs low pricing with a ~44% FrontierCode 1.1 score versus ~43% for Claude Sonnet 5 and ~41% for GPT-5.6 Terra.
2
$Meta Platforms, Inc.(META)$ has joined the HBF consortium alongside $SanDisk Corp.(SNDK)$ and $SK hynix(SKHY)$ as the first High Bandwidth Flash spec is released.
Meta’s involvement gives SanDisk another hyperscaler partner with a direct incentive to reduce HBM dependence and lower AI inference costs.
3
$SanDisk Corp.(SNDK)$ says HBF matches HBM’s 12.8 TB/s bandwidth while offering 4TB per GPU versus just 192GB dramatically expanding memory capacity for inference.
In internal Qwen3 testing, one HBF GPU handled a workload that needed eight HBM GPUs while four HBF GPUs matched the output of eight implying ~8x capex efficiency and ~2x GPU efficiency.
If that performance holds at scale then HBF could lower the number of GPUs required for inference and move NAND into a much higher-value layer of the AI memory stack.
$SanDisk Corp.(SNDK)$ is targeting mid-to-high teens growth with ~80% gross margins and ~50% FCF margins by FY28–FY30 while returning all excess cash to shareholders.
Sandisk says eight customers already represent ~$94B in lifetime contract value giving it massive visibility into that growth.
$SanDisk Corp.(SNDK)$ expects KV cache to drive 35% of AI data center NAND workloads by 2030 as longer context windows and larger batch sizes make cached inference data increasingly massive.
This is really interesting because KV cache sits in HBM today but because much of its written once and read repeatedly through techniques like prefix caching so NAND can handle a growing share without wasting premium HBM capacity.
If that shift plays out then NAND moves straight into the active AI memory hierarchy and gives SanDisk exposure to a much higher-value inference workload.
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.
