🎁AI Infrastructure Rally! AMD PTs Lifted, Lumentum & CoreWeave ERs Comes— What's Your Pick?

WallStreet_Tiger
08-10
Reward Tiger-CoinsReward 500 Tiger-coins

[Smile]Hey Tigers!

This week, the AI infrastructure space is hitting a triple-header. $Advanced Micro Devices(AMD)$ just dropped its Q2 scorecard on Aug 4 — revenue up 50%, data center doubling — and despite a short-term pullback, Wall Street has lifted price targets to ~$630.

Meanwhile, $Lumentum(LITE)$ and $CoreWeave, Inc.(CRWV)$ are both reporting after the bell tonight (Aug 11 ET / Aug 12 SG time).

One is the optical comms underdog; the other, the GPU cloud upstart. Together with AMD, they cover the compute layer, interconnect layer, and cloud layer of the AI infra stack — a full value chain.

1. AMD: Numbers Won, Valuation Resetting

AMD's Q2 was a blowout across the board. Revenue hit $11.54B, up 50% YoY — a record. Data Center revenue: $6.7B, up 107%, now 58% of total sales. Client segment: $3.1B, up 23%. Adjusted EPS: $1.66, beating the $1.62 consensus. Q3 revenue guide: ~$13B, up 41% YoY.

Yet the stock sold off ~7% post-earnings, from $518 to ~$483. Why? The Q3 guide, while above street consensus, missed the most bullish buy-side whisper numbers. After a 100%+ YTD run, the market has zero tolerance for "merely in-line." Capex also spiked from $282M last year to $808M, triggering short-term free cash flow concerns.

$AMD last traded at $483.36, with after-hours at $484.43.

Wall Street's reaction is the real story.

  • Post-print, Wells Fargo raised PT from $615 to $700;

  • UBS stays Buy at $730;

  • Rosenblatt lifted from $665 to $700;

  • Barclays at $665; Goldman at $640;

  • BofA at $620. Even the cautious Morgan Stanley bumped from $410 to $465.

Consensus PT now sits around $630 — ~30% upside from here.

This "stock down, targets up" divergence tells you the Q2 selloff was about expectation management and crowded positioning, not fundamentals. Three catalysts matter most:

  • 1. Helios rack-scale AI systems start shipping in Q3, with initial customers including OpenAI, $Oracle(ORCL)$ ,and $Meta Platforms, Inc.(META)$ — ramping in Q4. This is AMD's strategic pivot from "selling chips" to "selling systems." Management said Helios demand is already tracking above initial volume forecasts, with supply chain locked in.

  • 2. 2027 data center revenue guidance — management sees it more than doubling, with server CPU growth >70% and AI revenue growth "well above 100%." CEO Lisa Su explicitly called the $3B 2027 Instinct revenue target a "low-end estimate."

  • 3. The CPU/GPU mix shift. Training: GPU-to-CPU ratio ~8:1. Inference: drops to 4:1. Agentic AI: further to 1:1. This structural shift pushes AMD's CPU business into a market far larger than GPUs alone. Management estimates the server CPU TAM hits $220B by 2030.

Risks: Customer concentration is the key concern. AMD's AI revenue is heavily dependent on OpenAI, Anthropic, Meta, and Microsoft — unprofitable frontier labs, with Microsoft sitting at the center of both OpenAI funding and Anthropic cloud hosting. Memory cost inflation will pressure PC sales in coming quarters. And NVIDIA's CUDA moat remains formidable; AMD's ROCm catch-up is a multi-year game.

2. Lumentum: Optical Supply-Demand Tailwind — Can Q4 Make It Four Straight Beats?

$Lumentum(LITE)$ builds the fiber-optic nervous system inside and between AI data centers. Its EML laser chips, pump lasers, narrow-linewidth components, and Optical Circuit Switches (OCS) determine how fast data moves across GPU clusters. As AI clusters scale from 10K to 100K to 1M+ GPUs, optical interconnect bandwidth has become a tighter constraint than raw compute.

Lumentum reports FY2026 Q4 after the close on Aug 11. Guidance: revenue $960M–$1.01B, non-GAAP EPS $2.85–$3.05. Consensus: $988.6M revenue (+106% YoY), $2.99 EPS. Compare that to $0.88 EPS a year ago — this is a step-function jump. Lumentum has beaten estimates for four straight quarters, with an average beat of 9.6%.

$Lumentum(LITE)$ last traded at $890.17, up 6.22% on the day, with after-hours at $912.56. The stock is up 127% YTD but still ~18% below its 52-week high of $1,085.68.

tipranks.com

Pre-print analyst sentiment: Mizuho reiterated Outperform with a $1,100 PT, explicitly saying they'd buy ahead of earnings. Barclays upgraded from Equal-Weight to Overweight in July, PT $1,000. Rosenblatt has the Street-high at $1,300. Overall, 14–16 analysts rate it Buy.

public.com

Four key things to watch:

  • 1. 1.6T optical module shipment timeline. This is Lumentum's biggest growth driver. 1.6T products use 200G EML chips at roughly 2x the ASP of 800G. Management last quarter said 1.6T had passed qualification at key customers, with Q4 being the first quarter of meaningful revenue contribution. Any update on shipment volumes, customer purchase commitments, or capacity ramp will directly impact FY2027 revenue estimates.

  • 2. EML supply gap. Current EML chip supply shortage exceeds 30%; pump laser shortage is even worse, forcing customers into allocation mode. The gap caps near-term revenue upside but also means pricing power sits squarely with Lumentum. If management reveals capacity expansion timelines and capex plans, that's incremental alpha.

  • 3. OCS customer expansion. Lumentum's Optical Circuit Switches are already deployed at $Alphabet(GOOG)$ , $Microsoft(MSFT)$ , and $Amazon.com(AMZN)$ data centers — dynamically switching optical paths to improve GPU cluster utilization. A new cloud customer adopting OCS would significantly expand this segment's TAM.

  • 4. Q1 FY2027 implied guide. Mizuho expects Q1 revenue to grow >35% QoQ to ~$1.13B. If management guides near or above $1.1B, it confirms the second-half acceleration narrative.

Risks: The stock trades at ~10.9x forward P/S — well above industry average. TD Cowen in July cut its PT from $995 to $800 (Hold), arguing risk/reward has balanced out after the run-up. Also, Lumentum was added to the $NASDAQ 100(NDX)$ in May, forcing passive index inflows over the past three months — some of the gain may be flow-driven, not fundamental. EML and pump laser capacity expansion takes time; ramp execution is the key variable.

3. CoreWeave: The GPU Cloud Upstart's Moment of Truth — $99B Backlog on Trial

If AMD is the arms dealer, $CoreWeave, Inc.(CRWV)$ is the arms lessor — buying NVIDIA GPUs by the truckload, building data centers, and renting compute to AI companies on demand. In a market where AI compute is structurally undersupplied, CoreWeave sits at the critical middle layer between chipmakers and end users.

CoreWeave also reports Q2 after the close on Aug 11. Consensus revenue: $2.53–$2.56B, up 108–111% YoY — near the high end of guidance ($2.45–$2.60B). But EPS is expected at a loss of $1.17–$1.22, widening from $0.27 loss a year ago. CoreWeave is still in heavy investment mode: FY2026 capex guide is a staggering $31–35B, with Q2 alone expected at $7–9B.

$CRWV last at $90.67, up 6.26%, with after-hours at $91.92. The stock is up ~27% YTD but ~40% below its 52-week high of $153.20. Options are pricing in a 13–15.5% post-earnings move in either direction.

CoreWeave revenue by quarter (top panel) updated through the May 7, 2026 print, alongside Q2 2026 guidance midpoint as a hatched grey bar; financing events on the bottom panel with cumulative capital…来源:dds.finance

Three things that matter:

  • 1. Can the $99B backlog (RPO) keep growing? This number is CoreWeave's core visibility metric — far more important than quarterly P&L.

  • 2. Margin trajectory. Q1 operating margin was just 1%; adjusted EBITDA margin was 56%. With heavy depreciation and interest expense, any EBITDA margin compression will test market patience fast.

  • 3. Capacity execution. CoreWeave currently has >1 GW of active power capacity, targeting 1.7 GW by year-end and >8 GW by 2030.

The unique risk is the balance sheet. $31–35B annual capex means massive debt financing needs. Q2 interest expense alone is estimated at $650–730M. If rates stay elevated or funding conditions tighten, CoreWeave's cost of capital will directly erode already-thin margins.

Wall Street is "cautiously bullish": 15 Buy, 8 Hold, 1 Sell. Consensus PT: $128–137, implying 43–50% upside. BofA at $140; Deutsche Bank at $150; Wells Fargo at $155; Cantor Fitzgerald at $167. Barclays is the cautious voice: Equal-Weight, PT $90.

4. Three Tracks, One Tailwind

Put AMD, Lumentum, and CoreWeave together, and the AI infrastructure investment map comes into focus. All three ride the same mega-trend: AI compute demand keeps climbing, data center interconnect bandwidth doubles every ~2 years, and GPU cloud rental penetration is rising fast. But each faces a different flavor of risk: AMD = competitive risk, Lumentum = execution risk, CoreWeave = financial risk.

For AI infrastructure investors, tonight (Aug 11 ET / Aug 12 SG time) is a dense information window. Lumentum and CoreWeave reporting simultaneously gives you cross-validation from the interconnect layer and the cloud layer. AMD, with Q2 in the rearview, now waits for Helios volume and the 2027 data center doubling target to play out in H2. Same tailwind — but the alpha sources are diverging.

Tigers, what's your take on these three AI infra plays? Is AMD a buy-the-dip opportunity after the pullback? Worth a pre-earnings bet on Lumentum or CoreWeave? Drop your thoughts below! 👇


5. 🎯 Share Your Thoughts & Win Prizes

🎁 Here's what's up for grabs:

🐯 5 Tiger Coins for every valid comment on this post.

🐯 Extra 10 Tiger Coins for the first 10 and last 10 Tigers with valid comments.

🐯 Extra 15 Tiger Coins for the top 5 most popular and high-quality comments.

(Note: Tiger Coin rewards are randomly selected to encourage genuine, high-quality sharing. Spam and malicious behavior will be disqualified.)

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.
Reward 500 Tiger-coinsDeadline to 08/20 17:59
The originator will select the best recovery and allocate Tiger coins before the reward ends
Reward-post

Comments

  • 苏36
    08-10
    苏36
    AI Infra: Which One Stands Out?

    $AMD$, $LITE$ and $CRWV$ represent three different layers of AI infrastructure: compute, optical connectivity and AI cloud.

    AMD’s Q2 was strong, with revenue up 50% YoY and Data Center revenue more than doubling. Despite the post-earnings pullback, the bigger story is whether Helios and its next-gen AI systems can drive another growth cycle.

    For Lumentum, the key is 1.6T optics, EML supply and OCS adoption. If these accelerate, the optical bottleneck could become a major earnings catalyst.

    CoreWeave offers the fastest growth, but also the biggest risk. Investors need to watch its $99B backlog, margins, CapEx and debt load closely.

    My take: AMD for long-term strength, LITE for the optical bottleneck, CRWV for high-growth/high-risk exposure.

    The AI boom is still expanding. The real question now is: who captures the most profit from it?

    @WallStreet_Tiger [龇牙]

  • 北极篂
    08-11
    北极篂
    所以如果让我排序,我会是:


    AMD:基本面最扎实,等待估值重新定价;
    Lumentum:成长弹性最大,但最怕业绩不及预期;
    CoreWeave:成长故事最猛,但资产负债表风险最高。


    我的核心判断是:AI行情真正进入下半场后,市场不会再奖励“讲故事”,而会奖励能够把资本开支真正转换成收入、毛利和自由现金流的公司。
  • 北极篂
    08-11
    北极篂
    CoreWeave则完全是另一种投资逻辑。990亿美元级别的订单积压非常漂亮,但我最担心的也是它:收入增长很快,资本开支更快。如果GPU需求持续强劲,这种重资产扩张可以创造巨大规模;但一旦融资成本上升、GPU利用率下降或者客户集中度出现问题,巨额利息和折旧会迅速放大利润压力。
  • 北极篂
    08-11
    北极篂
    Lumentum则是我认为最容易出现业绩弹性的环节。AI集群规模越来越大,真正的瓶颈已经不只是GPU算力,而是GPU之间如何高速、稳定、低延迟地传输数据。1.6T光互连如果进入规模化出货,意味着它可能成为下一阶段AI资本开支的重要受益者。但这里要特别注意:股价已经大幅上涨,所以市场需要看到的不只是beat,而是产能扩张、客户订单和下一季度指引同时向上。
  • 北极篂
    08-11
    北极篂
    AMD的问题不是基本面不好,而是市场预期已经太高。数据中心收入超过翻倍、整体收入保持50%增长,本身已经非常强,但股价仍然回调,说明现在市场要求的不是“超预期”,而是必须明显超过最乐观预期。我反而认为这次调整更像估值消化,真正决定下一阶段股价的是Helios出货以及2027年数据中心收入能否继续高速增长。如果AMD能够从卖GPU进一步升级到卖完整AI系统,估值逻辑确实会发生变化。
  • 北极篂
    08-11
    北极篂
    我的观点:AI基础设施三条线同时验证,真正值得关注的是“业绩兑现能力”


    我认为这次AMD、Lumentum和CoreWeave同时进入财报窗口,非常有意思,因为三家公司刚好对应AI基础设施的三个关键环节:计算、光互连和算力云。如果三家公司都能给出强劲指引,市场交易的就不再只是AI芯片,而是整个AI资本开支周期仍在加速。
Leave a comment
9
1