WhiteLine Daily, a WuBlockchain publication, said recent earnings from CoreWeave, Supermicro and Lumentum all point in the same direction: AI demand has not shown a clear slowdown, but the factors limiting how fast computing capacity can be deployed are shifting away from GPUs themselves and toward financing, power, cooling, networking and optical interconnects.
At this stage, the report said, the more useful question is not simply who has the most orders. It is who can shorten time-to-compute and turn orders into usable compute capacity faster.
SMCI: demand is strong, but margin is the next thing to watch
Supermicro, listed as SMCI, reported fourth-quarter revenue of $11.1 billion, up about 93% from a year earlier. That figure came in slightly below market expectations, but the company entered fiscal 2027 with record backlog, posted more than $60 billion in new orders during the fourth quarter, and guided for FY2027 revenue of $65 billion to $72 billion.
In WhiteLine Daily’s reading, demand is not the main issue right now. Management said some revenue was delayed not because GPUs or servers were unavailable, but because customer-side power, cooling and networking were not yet ready. That matches a pattern seen across recent AI infrastructure earnings: producing a server does not mean a data center is already in a position to run it.
The report said investors should now pay closer attention to profitability. SMCI’s Q4 gross margin recovered to 17.5%, well above the previous quarter’s 9.9%, but full-year FY2026 gross margin was still only 10.8%. If margins can stabilize while orders and revenue continue to rise, the quality of this growth cycle would look stronger.
For short-term trading, WhiteLine Daily said the area around $38 to $40 could act as resistance. Still, it argued that margin durability matters more than a single price level.
Lumentum: margins beat expectations as tight optical supply moves through the chain
Lumentum, trading as LITE, posted fourth-quarter revenue of $1.01 billion, up 109% year over year. Its non-GAAP gross margin rose to 50.4%. Pump laser shipments kept growing and remained close to effectively sold out even after capacity expansion, suggesting that shortages in AI optical interconnects are no longer visible only in orders. They are also starting to show up in margins.
From there, the report laid out two names to watch across the supply chain.
SIVE: downstream optionality
Sivers’ DFB laser has entered GlobalFoundries’ silicon photonics reference design, and Jabil plans to use its DFB laser in 1.6T pluggable modules. WhiteLine Daily said Lumentum’s confirmation of tight laser supply and demand is a positive validation for SIVE’s industry positioning, though the next step still depends on scaled production and order conversion.
IQE: upstream visibility
IQE sits further upstream in the InP epitaxy segment. The company has signed a multi-year InP epitaxial wafer supply agreement with Tower Semiconductor and also secured a $14 million multi-year production order tied to AI and data centers. Compared with SIVE, which is still waiting for larger-scale production to ramp, IQE currently has firmer fundamental confirmation, according to the report.
Capital is still in AI infrastructure, but it is moving with the bottlenecks
From SMCI to Lumentum, WhiteLine Daily said this earnings cycle is sending an increasingly consistent message: capital has not left AI infrastructure. It is moving along the bottlenecks. After GPUs, power, cooling, networking and optical interconnects are becoming the factors that determine how quickly a batch of compute capacity can actually go live.
By that logic, the names best positioned in the current trade are the ones that can shorten the time-to-compute curve.