To The Moon

AI hardware stocks have suffered a sharp valuation reset, but the latest earnings show that underlying infrastructure demand remains strong. CoreWeave’s revenue backlog reached $104.2 billion, while Super Micro guided for up to $72 billion in annual revenue. The key bottlenecks are increasingly power, cooling, networking and financing—not a lack of AI orders

AI Infrastructure Rebounds After Hours

U.S. stocks ended the latest session lower as investors remained cautious ahead of the July CPI report:

• S&P 500: −0.32%

• Nasdaq Composite: −0.60%

• Dow Jones: −0.34%

However, several AI infrastructure names rebounded after the close:

• $CoreWeave(CRWV)$: up more than 14% after hours

• $Super Micro Computer(SMCI)$: up around 7%

• $Lumentum(LITE)$: roughly flat despite revenue more than doubling year over year

These very different reactions reveal how the market is currently pricing AI stocks.

Strong growth alone is no longer enough. Investors are rewarding companies that deliver a meaningful upside surprise in orders, margins or forward guidance.

CoreWeave’s Backlog Reaches $104.2 Billion

CoreWeave reported second-quarter revenue of $2.575 billion, up approximately 112% from $1.212 billion a year earlier.

The most important figure was its revenue backlog.

CoreWeave ended June with approximately $104.2 billion in contracted revenue backlog, up from $99.4 billion in the previous quarter. That figure does not include more than $25 billion in additional customer commitments secured early in the third quarter. CoreWeave earnings release

The company also disclosed that:

• Active power capacity reached 1.5 gigawatts

• Contracted power capacity increased to approximately 3.7 gigawatts

• Near-term computing capacity remains largely sold out

• Its first NVIDIA Vera Rubin NVL72 systems have been powered on and validated

The message is clear: customer demand remains strong, and CoreWeave already has significant visibility into future revenue.

The challenge is delivering the capacity quickly enough.

A $104 Billion Backlog Is Not the Same as $104 Billion in Profit

CoreWeave’s backlog represents contracted future revenue, but that revenue still depends on data centers being completed, computing capacity becoming available and services being delivered.

To support the expansion, the company raised its 2026 capital-expenditure forecast from $31–35 billion to $35–39 billion. Second-quarter capital expenditures alone reached approximately $9.4 billion. Reuters

CoreWeave also reported:

• Net loss of $626 million

• Net interest expense of $640 million

• Adjusted net loss of $567 million

The bullish case is easy to understand: AI computing demand is strong, capacity is heavily committed and contract visibility is improving.

The risks are equally visible: enormous capital requirements, high financing costs and potential delays in data-center construction.

CRWV remains one of the highest-beta ways to trade AI infrastructure demand.

Super Micro’s Biggest Surprise Was Margin Recovery

Super Micro reported quarterly revenue of approximately $11 billion, nearly double the $5.8 billion recorded a year earlier, although it came in slightly below the market’s expectation of roughly $11.55 billion.

The stronger signal came from profitability.

Gross margin reached 17.5%, compared with 9.9% in the previous quarter, while adjusted earnings reached $1.70 per share.

Investors have been concerned that intense competition for large AI customers would continue squeezing Super Micro’s margins. The latest quarter temporarily eased that concern. Super Micro earnings release

The company also issued aggressive forward guidance:

• Next-quarter revenue: $14.5–15.5 billion

• Fiscal 2027 revenue: $65–72 billion

• Previous Wall Street consensus: approximately $52.5 billion

Super Micro said it added hundreds of enterprise customers, secured more than $60 billion in new orders and entered fiscal 2027 with a record backlog. Reuters

That combination—stronger margins and a major guidance increase—explains why SMCI rallied despite missing the quarterly revenue estimate.

The AI Bottleneck Is Shifting

Super Micro said some customer projects were delayed because power, cooling and networking infrastructure was not ready.

That commentary closely matches CoreWeave’s situation.

The AI infrastructure debate is gradually shifting from GPU availability toward a broader set of constraints:

• Sufficient power supply

• On-time data-center construction

• Cooling systems for high-density server racks

• High-speed networking and optical connectivity

• Access to affordable long-term financing

This provides a positive demand read-through for companies such as $NVIDIA(NVDA)$, $Vertiv(VRT)$, $Arista Networks(ANET)$ and the optical-networking supply chain.

However, strong industry demand will not lift every related stock at the same time.

Lumentum reported 109% year-over-year revenue growth and guided above expectations, yet its shares barely moved after hours. That reaction shows how demanding valuations have become across high-flying AI hardware names.

The market is trading the gap between results and expectations—not simply the highest revenue-growth rate.

Two Important Tests Still Ahead

The first test is the U.S. July CPI report.

If inflation comes in soft and Treasury yields decline, CRWV, SMCI and other recently corrected AI infrastructure stocks could receive additional valuation support.

A hotter-than-expected report, especially with oil prices remaining elevated, could renew pressure on high-multiple technology stocks.

The second test is $Coherent(COHR)$ earnings.

SMCI and CRWV have confirmed that demand for servers and computing capacity remains strong. Coherent now needs to show that this investment is translating into stronger demand for 800G, 1.6T and laser products—while also improving margins and guidance.

Tiger Radar’s View

The latest results suggest that the recent AI hardware selloff remains more closely related to valuation and positioning than to a collapse in industry demand.

Still, the risk profiles differ significantly:

• CRWV: Exceptional backlog visibility, but the greatest capital-spending and financing risk

• SMCI: Strong revenue guidance and improving margins, but execution depends on customer infrastructure being ready

• NVDA: Positioned at the center of AI computing demand, with greater business certainty but less speculative upside

• COHR and LITE: Strong optical demand, though current valuations require increasingly large earnings beats

I would not chase CRWV or SMCI purely because of their after-hours gains.

The next signals to watch are whether Treasury yields decline after CPI and whether both stocks can hold their earnings gaps during regular trading.

If CRWV and SMCI open higher but quickly reverse, the market is still focused on valuation, financing and execution risk.

If they hold their gains on strong volume—and COHR provides another strong outlook—the recent AI hardware correction may be moving into a recovery phase.

Today’s Poll

Which AI infrastructure opportunity looks most attractive after earnings?

A. CRWV — $104.2 billion backlog and direct exposure to AI computing demand
B. SMCI — accelerating server revenue and recovering margins
C. NVDA — stay with the core AI winner instead of chasing higher-beta names
D. Wait for CPI and COHR earnings before buying the dip

Are CRWV and SMCI signaling the start of an AI hardware recovery—or is this only a short-lived earnings rebound?

Disclaimer: This post is for market discussion only and does not constitute investment advice. AI infrastructure companies remain exposed to valuation, financing, customer concentration, construction delays, capital-spending requirements and interest-rate risks.