To The Moon

MAG7 Exits the Trough, Yet Valuations Stay Compressed

The S&P 500 is nearing fresh highs again, but the more important story may be inside the MAG-7. 

Despite the recent rebound, valuations across the group still look compressed relative to the last major AI-driven peak, even as AI demand continues to broaden and monetization is becoming easier to see. 

That leaves room for a more durable rerating if the next leg of earnings confirms that hyperscaler spending is finally translating into stronger cloud revenue, better pricing, and improving returns on capital.

AI Demand Is Still Accelerating, and Compute Is Getting Tighter

•Anthropic is locking in supply across every major lane. After unveiling Mythos in early April, Anthropic secured about 3.5GW of future Google/Broadcom-backed compute starting in 2027, while also signing a multi-year CoreWeave deal that begins adding capacity later this year. AWS remains a key part of that story. Anthropic said Amazon remains its primary cloud provider and training partner, with Claude running across AWS Trainium, Google TPUs, and Nvidia GPUs. 

•OpenAI is still ramping, not slowing. OpenAI has continued pushing new model releases, and its expanded AWS partnership includes a commitment to consume about 2GW of Trainium capacity over time, another sign that leading labs are still pre-buying massive compute. 

•Meta-CoreWeave reinforces the same demand signal. Meta signed a fresh $21 billion CoreWeave agreement on top of a prior $14.2 billion deal, extending through 2032 and giving it access to early Vera Rubin deployments. 

Next, the following steps continue to drive the cycle：

Better models and agents still drive more usage;

Token growth continues to expand exponentially;

GPU availability decreases while GPU cloud service usage rises;

The monetization picture is getting clearer.

Amazon has now disclosed that AWS AI services are running at more than $15 billion in annualized revenue, or roughly 10% of AWS, while its chip business has exceeded a $20 billion annualized revenue run-rate. Morgan Stanley also highlights AWS pricing increases, a backlog approaching $350 billion, and a path to 31% and 33% AWS growth in 2026 and 2027.

Valuations Still Look Compressed

The Mag-7 forward P/E sits at 24.9 times as of April 14, while the S&P 500 large-cap forward P/E is 20.5 times. For a market standing close to new highs, that is not an aggressive multiple. It looks more like a market that has already digested a lot of macro fear, capex skepticism, and rate pressure.

If we look within the MAG7, we find an even more interesting situation.

Following the recent strong rebound, the latest (as of Apr.14) forward P/Es for Microsoft (21x), Nvidia (22x), Amazon (25.8x), and Meta (19.5x) all remain at the lower end of their historical valuation ranges. 

Given the logic of improving CAPEX ROI expectations mentioned earlier, it is reasonable to conclude that these core assets are poised for further valuation rerating and catch-up plays.

The Best-Positioned Names

AMZN, NVDA, and GOOG still look like the cleanest beneficiaries of this phase.

$Amazon.com(AMZN)$   has the most direct line into rising AI spend through AWS, improving backlog visibility, instance pricing, and its own silicon stack. 

$Alphabet(GOOG)$   has the strongest “full-stack” upside, with search, cloud, TPUs, and Gemini all feeding into the same monetization engine, and Morgan Stanley explicitly argues Google Cloud could become 31% of company-wide EBIT by 2028.  

$NVIDIA(NVDA)$   remains the system-level beneficiary because even as custom silicon grows, many of the largest new capacity deals are still being built around Nvidia-heavy infrastructure. Reuters reported that Meta’s new $21 billion CoreWeave agreement gives it access to Vera Rubin-based systems, which is a reminder that the industry is still racing to secure premium AI capacity, not walking away from it.  

$Microsoft(MSFT)$   may have catch-up potential, but the stock could remain a second-order beneficiary in the near term. Bernstein is constructive on Azure growth, contract backlog, and valuation, and says Microsoft is strongly positioned even if the AI cycle becomes more volatile. The stock is still partially trapped by its broader software identity, which can mute the market’s willingness to pay peak AI-beta multiples compared with the more direct cloud and chip beneficiaries. 

@TigerStars  @CaptainTiger  @TigerWire  @Daily_Discussion  @Tiger_chat  @Tiger_comments  @MillionaireTiger