Goldman Sachs TMT Conference: SanDisk Sees Long-Term NAND Supply Constraints; AI Commercialization Advancing on Multiple Fronts
nashnova research
SanDisk told the Goldman Sachs tech conference that NAND supply will stay tight for the foreseeable future, while AI inference demand keeps climbing — a gap that could underpin storage prices; multiple companies at the same event disclosed AI is already cutting real costs and lifting efficiency.
Why would NAND supply stay tight long-term?
SanDisk's management stated clearly: growth in NAND (flash-memory chips — the core storage component in phones and servers) supply will likely remain constrained for the foreseeable future.
This means → chipmakers have no plans to ramp capacity aggressively; the supply side stays disciplined.
At the same time, the mass rollout of AI inference (running trained models to actually answer questions and generate content) is steadily pushing storage demand higher.
In plain terms = supply barely grows while AI pulls demand hard — the gap could widen, giving NAND prices long-term support.
How is AI actually making money on e-commerce and travel platforms?
Etsy defined AI as its core efficiency tool: AI now achieves "deeper inventory understanding, buyer understanding, and purchase-intent recognition," pushing personalized listings to users and lifting conversion rates.
Etsy is also partnering with third-party LLMs and emerging AI platforms so its differentiated inventory surfaces at every shopping entry point; management said traffic from AI-agent platforms is still small but carries "extremely high purchase intent."
Booking.com is working closely with hyperscale cloud providers and generative-AI companies to test new consumer-interaction models; management expects these channels to evolve into paid performance-marketing channels, but LLM-driven traffic currently accounts for less than 1% of total traffic.
On the cost side, Booking.com is already applying generative AI in customer service and software development, reporting it is "seeing real benefits."
Can AI really let companies cut staff without losing output?
CHYM disclosed a 10% headcount reduction aimed at building "a flatter, faster organization"; its product and engineering teams shrank to half their original size, yet output stayed flat because "accountability is clearer and operations are leaner."
This means → AI as an efficiency tool is no longer just a talking point — at least at this company, halving the team while maintaining output is already reality.
Block, after a major organizational restructuring at the start of the year, is achieving significant operating leverage on a fixed cost base; a unified AI infrastructure built over the past two years has "meaningfully accelerated product iteration and system stability."
Are consumers still spending?
Block management said consumer spending showed exceptional resilience throughout Q2, and that strength continued into July and August.
Toast said spending levels remain stable — "nothing new to call out since earnings" — consistent with Visa's earlier remarks at the conference that consumer trends are steady.
Booking.com noted that travel demand remains resilient despite ongoing macro and geopolitical uncertainty, largely in line with its Q2 earnings commentary.
Are there any weak signals?
Comcast flagged that softness in the Orlando theme-park market, which emerged in June, has carried into Q3.
Management attributed it to a possible combination of macro headwinds and a pull-forward effect from Epic Universe's opening last year.
This reflects that consumer resilience is not monolithic — offline experiential spending is already showing cracks in pockets.
What to watch next?
SanDisk's medium-to-long-term NAND supply call, combined with the ongoing expansion of AI inference demand, forms the support framework for storage prices.
This means → the key test ahead is whether the supply side truly stays disciplined and resists ramping when prices rise.
On AI monetization, companies have moved from "telling the story" to "showing the math" — traffic share, headcount cuts, and cost savings now come with real numbers. The next question is whether those numbers keep scaling.
市场有风险，内容仅供研究参考，不构成投资建议。