Top Companies' AI Moves

Today's Stories

1. 1

   Oracle cloud sales jump 121% to $7.4 billion, topping estimates

   Oracle said Thursday that sales in its closely watched cloud infrastructure business jumped 121% to $7.4 billion, above the $7.19 billion analysts expected, and added 850 megawatts of data center capacity. The cloud unit is now the engine of Oracle's results, with total fiscal first-quarter sales up 30% to $19.3 billion and profit of $1.92 a share versus the $1.75 analysts expected.

   Oracle completed its earlier-announced plan to sell $20 billion of equity at market value, so the test is whether the build-out for OpenAI and other customers keeps pace with that spending.

2. 2

   Apple's SimpleDesign model trains on 2M protein pairs, skips multi-stage

   Apple researchers presented SimpleDesign, a multi-modal protein design model trained directly in data space with a single-stage objective, combining cross-entropy for sequences and regression for structures. It was trained on over 2M sequence-structure pairs. Existing models typically rely on a multi-stage training process with autoencoders, but SimpleDesign shows this is not necessary for competitive co-design performance on benchmarks.

   The approach hinges on whether single-stage training can match multi-stage models on broader protein engineering tasks. Watch for adoption in drug discovery workflows.

3. 3

   Marvell targets $18 billion FY2028 revenue, lifts outlook to $30 billion

   Marvell Technology is targeting $18 billion in FY2028 revenue, raising its combined FY2027 and FY2028 expectations from roughly $20 billion to $30 billion, per the analysis. Connectivity, switching, and storage drove most of the FY2028 forecast increase, reducing Marvell's dependence on individual custom silicon programs, the article says.

   The test is whether Marvell converts its over 20 XPU and XPU-attach design wins into FY2029 custom ASIC revenue of $10–$11 billion. Watch the 42.5x forward EV/EBITDA multiple.

4. 4

   The Cognition Index Research Report: 62% want human intuition over AI

   Philip Morris International and WSJ Intelligence unveiled The Cognition Index Research Report, a survey of more than 2,500 business professionals in the US, UK, Italy, South Africa and Brazil. 62% say human intuition should take priority over AI in creative work, and 50% say the same for market analysis, meaning professionals still want people deciding even where data drives the work.

   Only 25% strongly agree their organization has a clear, effective process for fact-checking AI outputs, so the test is whether firms build verification frameworks as AI enters workflows.

5. 5

   Nvidia plans major data centre expansion in Australia to meet AI demand

   Reuters reports that Nvidia plans a major expansion of data centre capacity in Australia to meet AI demand. The plan signals Nvidia's push beyond selling AI chips into building data centre capacity in Australia.

   The scale and timeline of the expansion will hinge on further details beyond Reuters' report, which stays unspecified here.

6. 6

   Lam Research: chip equipment "fundamentally sold out"

   At a Goldman Sachs event, an unnamed Lam executive said the semiconductor equipment industry is "fundamentally sold out" and constrained by clean-room capacity. Lam sees eight to 10 new clean rooms at tier-one customers coming online between now and the end of next year. Lam said its immediate concern is missing upside demand, not preparing for a downturn, even as it acknowledges the industry's historical cyclicality. Its Customer Support Business Group posted a third consecutive record revenue quarter at nearly $2.5 billion.

   Lam's bet on new capacity hinges on whether NAND conversion—about $40 billion of spending through the end of next year—actually forces new capacity as wafer starts fall. Watch its 52% gross margin, its highest in 20 years, against guidance of 51% to 52%.