technologybriefs
9:30in productionCh. 1 · Origin/ 9:30 · ceiling 15 min
Semiconductors · Hardware

Mellanox Technologies

1999

Mellanox didn’t invent speed—it weaponised standardisation against latency, then vanished into Nvidia before its full Ethernet promise could be tested.

Mellanox Technologies was an Israeli-American fabless semiconductor company founded in May 1999 in Yokneam Illit, with financial offices in Santa Clara. It specialised in high-speed, low-latency interconnects—primarily InfiniBand and Ethernet—delivering adapters, switches, software, cables and silicon for HPC, data centres, cloud, storage and financial services. Its products enabled RDMA and RDMA over Converged Ethernet. By 2011, its InfiniBand solutions were embedded in many TOP500 supercomputers. It expanded into Ethernet-based storage area networks from 2011. A $64 million venture round in 2002—backed by Intel, IBM, Sequoia Capital and U.S. Venture Partners—confirmed early market confidence. Nvidia acquired Mellanox for $6.9 billion in March 2019; the deal closed in April 2020 after antitrust approvals. Mellanox was integrated into Nvidia’s networking division, and the brand was discontinued for new products.

SemiconductorsHardwareSemiconductorsHardwareComputingSemiconductors across the network →
Chapters & takeaways4
  1. 1:04
    Origin

    Founded in May 1999 in Yokneam Illit by ex-Intel and Galileo executives, with financial offices in Santa Clara.

  2. 2:48
    Markets & Products

    It sold adapters, switches, software, cables and silicon for HPC, data centres, cloud, storage and finance—based on InfiniBand and Ethernet.

  3. 4:10
    Early Validation

    A $64 million VC round in 2002—led by Intel and IBM—validated its technical path before first-gen chips shipped.

  4. 5:20
    Exit

    Acquired by Nvidia for $6.9 billion in 2019; fully integrated and rebranded by April 2020.

Worth your time?

Yes. Study the whole thing.

4/ 5
What works
  • multi-protocol interconnects up to 200 Gbit/s
  • RDMA acceleration in Ethernet environments
  • TOP500 deployment at scale by 2011
What does not
  • scale its Palestinian engineering programme beyond reported headcount
  • maintain independent product development post-2020
Study it if
  • HPC infrastructure engineers
  • networking hardware architects
  • historians of semiconductor acquisitions
Skip it if
  • cloud platform developers relying solely on post-2020 Nvidia networking docs
  • startups seeking open Mellanox SDKs or roadmaps
The written brief1 min read

What it is and the problem it solves

Mellanox was a fabless Israeli-American semiconductor company founded in May 1999 to solve latency and bandwidth bottlenecks in HPC, cloud, and financial systems using InfiniBand and Ethernet switched fabrics.

How it works

Mellanox built fabless semiconductor chips—fabricated by TSMC—that powered adapters, switches, cables, and software for InfiniBand and Ethernet networks. Its products enabled RDMA and RDMA over Converged Ethernet, delivering high data rates and low latency across switched fabrics.

What works

Its ConnectX, Quantum, Spectrum, and LinkX families delivered interoperable, high-speed interconnects up to 200 Gbit/s. By 2011, its InfiniBand products ran in many TOP500 supercomputers. Its expansion into Ethernet-based storage area networks from 2011 onward succeeded commercially and technically.

What does not

It did not sustain independent product development beyond 2020. It did not scale its Palestinian engineering engagement beyond ‘over 100’ by 2018, nor did it publicly report outcomes of that initiative.

What it changes

It shifted high-performance computing from proprietary interconnects toward standardised, multi-protocol fabrics. It helped mainstream RDMA outside InfiniBand—especially in Ethernet-based storage—and accelerated adoption of 100+ Gbit/s interconnects before 2020.

Is it worth your time

Yes—if you work on high-performance interconnects, storage area networks, or HPC infrastructure pre-2020. Post-acquisition, its IP lives inside Nvidia’s networking stack, but the standalone Mellanox brand and roadmap ended in 2020.

Same field · Semiconductors4 of 51
9:51
Hardware for artificial intelligenceAI hardware is a set of purpose-built chips—Lisp machines, GPUs, TPUs, NPUs—that accelerate AI workloads by optimising for parallelism, memory bandwidth, and low-precision arithmetic. It delivers real speed and efficiency gains, but at the cost of flexibility, portability, and architectural transparency. Its dominance reflects engineering necessity—not inevitability.
9:20
ChipletChiplets are modular ICs—functional fragments designed for assembly on an interposer. They enable heterogeneous integration, pre-test validation, and IP reuse. But they shift complexity from fabrication to integration, and require standards compliance to deliver on their promise.
9:27
ECC memoryECC memory adds redundancy and decoding logic to catch and fix single-bit memory errors before they propagate. It works via SECDED Hamming or Hsiao codes, using 8 extra bits per word—64+8 for DDR–DDR4, 32+8 per subchannel for DDR5. Mandatory on-die ECC in DDR5 and LPDDR6 addresses rising soft-error rates from shrinking process nodes. It delivers deterministic single-bit correction in servers and critical infrastructure—but offers no protection beyond that, and no benefit where silent corruption is tolerable.
9:33
EUV lithographyEUV lithography replaces transmissive deep-UV optics with reflective Mo/Si mirrors, tin-plasma light sources, and vacuum operation to pattern sub-7 nm semiconductor features. It works — but only after decades of co-development across continents, and only where photon budget, thermal drift and stochastic noise can be managed. It is necessary, not optional, for leading-edge logic — but it does not generalise, simplify or cheapen.
Up next in Technology

Memcached

Brad Fitzpatrick · 2003 · 9:13

Memcached doesn’t make your app faster—it makes your database quieter.

Play nextBack to Semiconductors
9:13