• NVIDIA (NVDA) is the preferred large-cap AI chip stock on relative valuation; consider it ahead of AMD and Broadcom, with the upcoming earnings report as the near-term catalyst.
• Micron (MU) stands out as a low-valuation way to invest in AI memory demand, though its outlook depends on continued data-center investment.
• For higher-risk AI data-center exposure, the speaker favors IREN and Nebius (NBIS) as inexpensive NeoCloud names; weigh Texas policy uncertainty, especially for future projects.

NVIDIA (NVDA)

• The speaker considers NVIDIA the best value among the large AI chipmakers, citing an EV/GP/RG figure of 0.3.
• They describe NVIDIA as roughly twice as cheap as AMD and about one-third cheaper than Broadcom by this measure.
• They view AMD’s strong earnings report as a positive signal ahead of NVIDIA’s earnings, which they said were due the following week.
• The speaker was surprised NVIDIA had fallen about 5% over the prior 12 months, despite the company having nearly doubled in size.
• NVIDIA’s upcoming earnings were identified as a potential catalyst for an AI-stock rebound.

Takeaways

• Within the speaker’s valuation framework, NVDA is their preferred large-cap AI chip stock over AMD and Broadcom.
• NVIDIA’s next earnings report was the key near-term event in the discussion; the transcript offers no price target.

Advanced Micro Devices (AMD)

• The speaker called AMD’s latest earnings report “blockbuster” and “absolutely outstanding,” but noted the stock was up only 1% after giving back most of its gains.
• Despite the results, the speaker considers AMD less attractive than NVIDIA because AMD’s EV/GP/RG valuation is roughly twice NVIDIA’s.

Takeaways

• The speaker views AMD’s results as evidence of strong AI demand, but favors NVDA on relative valuation.
• The discussion does not include a specific recommendation to buy or sell AMD.

Broadcom (AVGO)

• Broadcom was down about 2% on the day discussed.
• The speaker considers Broadcom roughly one-third more expensive than NVIDIA by their EV/GP/RG measure and says they have difficulty justifying AMD or Broadcom at those relative valuations.

Takeaways

• The speaker’s comparison favors NVDA over AVGO on valuation.
• No price target or specific timing was given for Broadcom.

Micron Technology (MU)

• Micron fell 7% on the day discussed, while the speaker argued that demand for high-bandwidth memory and DRAM used in AI data centers should continue.
• The speaker said Micron’s EV/GP/RG figure was 0.07, calling it the cheapest AI stock outside the NeoCloud group.
• They argued that data-center construction could continue in the U.S. or move abroad, and potentially into space in the longer term.
• The broader memory group was down about 5% that day.

Takeaways

• The speaker sees MU as a particularly inexpensive way to gain exposure to AI-related memory demand.
• Their thesis depends on continued data-center and GPU-rack investment; the transcript does not quantify how changes in construction would affect Micron’s business.

SK Hynix

• The speaker said SK Hynix has a valuation similar to Micron’s, but that its U.S.-traded ADR carries a premium.
• They noted that a memory-sector ETF can hold SK Hynix’s Korean-listed shares, potentially avoiding the ADR premium.

Takeaways

• For investors interested in memory exposure, the speaker suggested considering a sector ETF rather than paying the ADR premium directly.
• No specific ETF or ticker was named.

Memory-sector ETF

• The speaker said memory stocks appeared inexpensive across the board and suggested an ETF as a way to invest in the sector.
• They highlighted that a sector ETF could hold SK Hynix’s Korean listing rather than its U.S. ADR.

Takeaways

• A memory ETF may offer diversified exposure to the sector and, in the speaker’s view, avoid paying the ADR premium on SK Hynix.
• The transcript does not name a specific fund or discuss its fees or holdings in detail.

Nebius (NBIS)

• The speaker described Nebius as “super cheap,” with the broader NeoCloud group trading below 0.1 on their valuation measure.
• They argued that Nebius’s international operations make it less exposed to Texas data-center policy concerns than the market appeared to assume.
• The speaker said Nebius was building and operating data centers in multiple locations, including in the U.S.

Takeaways

• The speaker views NBIS as undervalued relative to its data-center growth and international footprint.
• The discussion highlights perceived market concern about data-center restrictions, but the speaker believes Nebius is less exposed to the Texas issue.

IREN

• IREN fell about 11% on the day discussed. The speaker called it the cheapest NeoCloud stock on their valuation measure.
• They said IREN’s Texas operations were grandfathered in and that projects already in progress would not be affected by a possible new moratorium.
• The speaker argued that the market was overreacting to the possibility of a broad data-center ban.

Takeaways

• The speaker sees IREN as an especially low-valued NeoCloud, while emphasizing that existing Texas projects should be less exposed to proposed restrictions.
• Texas data-center policy remains a risk discussed in the transcript, particularly for future projects.

SpaceX (SPCX, as labeled in the episode title)

• The speaker grouped SpaceX with NeoCloud companies, saying that most of its revenue over the next year would come from NeoCloud.
• They described it as one of the cheapest names in the group and said its data centers were in Tennessee, not Texas.
• The transcript treats SpaceX as an investment name but does not discuss how an investor could access it or establish that it is publicly traded.

Takeaways

• The speaker’s valuation view is favorable, based on expected NeoCloud revenue and the location of its data centers.
• The transcript provides no price target or details on public-market availability.

CoreWeave (CRWV)

• CoreWeave was described as “super cheap,” with the NeoCloud group trading below 0.1 on the speaker’s valuation measure.
• The speaker said the group’s share-price declines appeared disconnected from fundamentals.

Takeaways

• The speaker sees CRWV as inexpensive within the NeoCloud theme, but offers no company-specific operating details or price target in this discussion.

Credo Technology (CRDO)

• Credo was presented as a connectivity play focused more on copper than fiber optics, though the speaker said it overlaps with Applied Optoelectronics.
• The speaker cited expected sales growth of more than 100% over the next 12 months, an EV/GP/RG figure of 0.22, and a Rule of 40 above 120.
• Despite those figures, the speaker said Credo looked expensive relative to NVIDIA and other opportunities.

Takeaways

• The transcript points to strong projected growth in AI connectivity, but the speaker prefers NVDA’s relative valuation to CRDO’s.
• The growth figure is a forecast cited by the speaker, not a guaranteed outcome.

Applied Optoelectronics (AAOI)

• AAOI was described as a fiber-optics connectivity company, with some overlap with Credo.
• The speaker cited expected sales growth of 120% over the next 12 months, an EV/GP/RG figure of 0.23, and a Rule of 40 above 120.
• AAOI fell about 7% on the day discussed. The speaker considered its valuation less compelling than NVIDIA’s.

Takeaways

• AAOI offers exposure to AI-related fiber-optic connectivity, with very high growth expectations cited in the transcript.
• The speaker’s comparison suggests looking at NVDA instead if relative valuation is the priority.

Marvell Technology (MRVL)

• Marvell was mentioned as a company in the “beyond memory” AI infrastructure group.
• The speaker was not persuaded by the valuation of Marvell relative to NVIDIA and the lower valuation figures cited for Credo and AAOI.

Takeaways

• The speaker did not express a positive valuation view on MRVL compared with NVIDIA.
• No specific valuation figure or recommendation was provided.

Lumentum (LITE)

• Lumentum was mentioned as another “beyond memory” AI infrastructure company.
• The speaker cited an EV/GP/RG figure of 0.61 and said that valuation made them prefer NVIDIA.

Takeaways

• Within the speaker’s comparison, LITE was less attractive than NVDA on valuation.
• No price target or specific recommendation was given.

Astera Labs (ALAB)

• Astera Labs was included among the AI infrastructure companies the speaker compared with Credo and AAOI.
• The speaker was not compelled by its valuation relative to NVIDIA, but gave no specific valuation figure.

Takeaways

• The speaker’s comments suggest a preference for NVIDIA over ALAB based on relative valuation.
• The transcript provides no company-specific growth forecast or price target for ALAB.

AppLovin (APP)

• AppLovin was mentioned as an example of what the speaker considered highly unpredictable market pricing: it was up about 35% while AI infrastructure stocks were falling.
• The speaker did not present an investment thesis for AppLovin.

Takeaways

• AppLovin was used as a comparison to illustrate the speaker’s view that recent valuations and price moves were disconnected from fundamentals, not as a buy or sell idea.

AI infrastructure and data-center sector

• The speaker described continued fear around AI valuations, with semiconductor, memory, NeoCloud, and connectivity stocks falling even after strong company results.
• They argued that proposed data-center restrictions did not amount to a blanket ban and would have less effect on projects already underway.
• The speaker identified NVIDIA’s next earnings report as the main near-term catalyst they were watching.

Takeaways

• The discussion favors AI infrastructure exposure, especially memory and NVIDIA, on the view that demand and data-center construction will continue.
• The transcript also identifies policy uncertainty around data-center construction—particularly future projects in Texas—as a risk, while arguing that the market may be overestimating its impact.