The questions · AI infrastructure · Where is the order backlog building in AI infrastructure?

Where is the order backlog building in AI infrastructure?

Backlogs, order intake and bookings by segment · 194 findings in the last 90 days · 519 since 5 Dec 2016 · latest 31 Oct 2026

Frontier Tech Intelligence reads 100+ sources in eleven languages every day and tags every dated finding that bears on this question. In the last 90 days, 194 findings in AI infrastructure did; the companies named most often were Anthropic (12), NVIDIA (12), Meta (9). Since 5 Dec 2016 the count is 519. Below are the latest findings with their outlet and date, and the companies they name.

The latest findings

  1. Megmeet is advancing first customer projects for Vera Rubin-related products.

    Megmeet said its Vera Rubin-related products are being launched and connected with customers in parallel with industry peers, and that it is advancing orders for its first projects with the goal of early deployment. Megmeet expects GB300 and Vera Rubin solutions to coexist for an extended period rather than one immediately replacing the other.

    36Kr newsflashes

  2. OpenAI spent about $10 million on a Navier-Stokes proof and hundreds of millions of dollars on math research.

    OpenAI spent on the order of $10 million for a Navier-Stokes proof and less than that on the average proof, with total mathematical research spend in the hundreds of millions of dollars, including unpublished discoveries.

    Peter H. Diamandis · OpenAI

  3. Privacy-focused AI startup Ollie is shutting down.

    Ollie, a U.S. startup behind a family-focused app for booking appointments, planning meals, and paying bills, is shutting down.

    Tech in Asia

  4. AWS revenue accelerated 39.3% year over year, with its order backlog reaching $578 billion.

    Amazon Web Services posted 39.3% year-over-year revenue growth in the July–September 2026 quarter, approximately its fastest growth in five years. AWS’s order backlog reached $578 billion.

    ITmedia News · Amazon

  5. SanDisk has locked roughly half of its fiscal 2027 capacity to NBM projects.

    SanDisk said NBM projects account for approximately 50% of its capacity in fiscal 2027, rising to about two-thirds in fiscal 2028. Management expects the NBM backlog to exceed the company’s available supply by the end of the decade.

    36Kr newsflashes · Sandisk

  6. Lasertec’s FY2027 order outlook is ¥400–500 billion despite a weak Q1.

    Lasertec’s higher expected FY2027 orders of ¥400–500 billion are projected to offset weakness in the first quarter, indicating sustained demand for its semiconductor inspection equipment.

    TradingView · Lasertec

  7. LG CNS is expanding AI-agent work within banking system-integration projects.

    LG CNS reported that banks are placing separate orders for AI agents and adding AI functions to ongoing system-construction projects, creating additional development demand alongside existing SI work.

    ZDNet Korea · LG CNS

  8. GE Vernova’s Q3 2026 orders nearly doubled to $29.3 billion.

    GE Vernova reported Q3 2026 orders of $29.3 billion on October 28, 2026, up 96% year over year, with a book-to-bill ratio above two. Its total backlog reached $197 billion, including $104 billion in equipment backlog and $93 billion in services backlog.

    AIM (Analytics India Magazine) · GE Vernova

  9. American Electric Power’s contracted large loads rose 22% to 84 GW, driven mainly by an Ohio data center.

    The increase was primarily linked to a 10.4-GW data center in Ohio being developed by SB Energy.

    Utility Dive · American Electric Power, SB Energy

  10. Anthropic has more than $500 billion in compute commitments, including $120.5 billion with SpaceX.

    Anthropic’s compute commitments exceed $500 billion, with much of the contracted capacity described as non-cancellable. The commitments include a $120.5 billion agreement with SpaceX.

    Peter H. Diamandis · Anthropic, SpaceX

  11. NVIDIA’s chip order book is sold out for approximately two years.

    NVIDIA chips are reportedly unavailable to new buyers for two years, reflecting sustained demand and a fully booked order pipeline.

    Moonshots with Peter Diamandis · NVIDIA

  12. Mark Cuban Cost Plus Drug Company developed tractor-trailer-sized modular pharmaceutical manufacturing pods.

    The company has packaged the same AI and robotics capabilities into pods roughly the size of a tractor trailer that can be shipped and configured to produce different drug types. The U.S. government was its first customer; the company recently shipped the first pod and received an order for a second, with potential deployment for disaster-response and military use.

    a16z

Members see all 194 findings of the period and the answer written from them, by segment and company, updated daily.

Named most often, last 90 days

12Anthropic12NVIDIA9Meta8OpenAI7Alphabet6Amazon6Microsoft5CoreWeave

Mentions in findings that bear on this question. Each company has a file.

The same question elsewhere

Where is the order backlog building in humanoid robotics?33 findings in 90 daysWhere is the order backlog building in robotics?49 findings in 90 daysWhere is the order backlog building in the space industry?95 findings in 90 days

Other questions in AI infrastructure

Where is the money going in AI infrastructure?5,232 findings in 90 daysWhich AI infrastructure suppliers are funded to deliver?3,097 findings in 90 daysWho is expanding production in AI infrastructure?2,668 findings in 90 daysWhat AI infrastructure capacity is coming online?1,786 findings in 90 daysHow much of AI infrastructure is China?1,431 findings in 90 daysWhat is blocking AI infrastructure?744 findings in 90 daysWhat went wrong in AI infrastructure?727 findings in 90 daysWho owns AI infrastructure: states, founders or funds?523 findings in 90 daysWhere is supply constrained in AI infrastructure?493 findings in 90 daysWhich AI infrastructure suppliers changed hands?492 findings in 90 daysWho is buying whom in AI infrastructure?489 findings in 90 daysWho raised or cut guidance in AI infrastructure?447 findings in 90 daysWho supplies whom in AI infrastructure?442 findings in 90 daysWho won qualifications and design wins in AI infrastructure?375 findings in 90 daysWhere are prices moving in AI infrastructure?281 findings in 90 daysWho controls what in AI infrastructure?261 findings in 90 daysWho is new in AI infrastructure?241 findings in 90 daysWho moved into mass production in AI infrastructure?82 findings in 90 daysWhere are second sources appearing in AI infrastructure?69 findings in 90 days

Questions this page answers

Where is the order backlog building in AI infrastructure?

194 dated findings in AI infrastructure bear on this in the last 90 days (backlogs, order intake and bookings by segment). The companies named most in the last 90 days: Anthropic (12), NVIDIA (12), Meta (9), OpenAI (8), Alphabet (7). Latest: Megmeet is advancing first customer projects for Vera Rubin-related products. (36Kr newsflashes, 31 Oct 2026) OpenAI spent about $10 million on a Navier-Stokes proof and hundreds of millions of dollars on math research. (Peter H. Diamandis, 31 Oct 2026) Privacy-focused AI startup Ollie is shutting down. (Tech in Asia, 31 Oct 2026) Frontier Tech Intelligence, from 100+ sources.

Which companies in AI infrastructure are named most for backlogs?

The companies named most in the last 90 days: Anthropic (12), NVIDIA (12), Meta (9), OpenAI (8), Alphabet (7). Counted over 194 findings of the last 90 days.

What are the latest findings on backlogs, order intake and bookings by segment in AI infrastructure?

Megmeet is advancing first customer projects for Vera Rubin-related products. (36Kr newsflashes, 31 Oct 2026) OpenAI spent about $10 million on a Navier-Stokes proof and hundreds of millions of dollars on math research. (Peter H. Diamandis, 31 Oct 2026) Privacy-focused AI startup Ollie is shutting down. (Tech in Asia, 31 Oct 2026)

Twenty standing questions, answered from the evidence every day.

Members read the answers written from all the findings, by segment and company, ask their own questions across 100,000+ signals and get the week explained. $99 a month, 7-day trial. What Frontier Tech Intelligence is →

Subscribe · $99 a month