Habana Labs

Acquired by Intel

Habana Labs

Habana Labs is an Israel-based AI chip company, acquired by Intel, that designs specialized processors and software stacks to make large‑scale deep learning training and inference more efficient and cost-effective for data centers and cloud providers. [nof3g7] [1srb2g] [7vx0xk]
Habana Labs is a for-profit semiconductor company that develops high-performance AI accelerators, notably the Intel Gaudi and Goya processor families, for deep learning training and inference workloads in data centers and cloud environments. [nof3g7] [1srb2g] [rtnhj5] Founded in 2016 and headquartered in Tel Aviv, Israel, the company focuses on custom AI processors and the SynapseAI software stack, and since 2019 has operated as an Intel subsidiary while retaining a distinct product line within Intel’s data-center portfolio. [1srb2g] [7vx0xk] [db0lbb] Consultants track Habana because its Gaudi platform is one of the main non‑GPU options for large-scale AI training, used in offerings like Amazon EC2 DL1 instances to reduce AI training cost. [qnw5d4] [rpzmf3]

Identity and Form

  • Legal form and jurisdiction:
    • Privately held (post-acquisition) semiconductor company, originally an independent startup, now an Intel subsidiary headquartered and incorporated in Israel; Intel “bought Habana Labs” in December 2019 and describes it as “Habana Labs (an Intel company)”. [on61d7] [db0lbb] [rpzmf3]
  • Headquarters and presence:
    • Headquarters in Tel Aviv, Israel, with operations focused on data-center and cloud AI hardware deployed globally via cloud providers and enterprise data centers. [1srb2g] [qnw5d4] [rtnhj5]
  • Size:
    • StartupHub.ai reports approximately 212 employees on record for Habana Labs. [1srb2g]
  • Where it lives online:
    • Homepage: (canonical domain inferred from context as the “Habana Labs” product/brand site under Intel; external profiles point to Habana as “now part of Intel” and do not list a separate public corporate site). [nof3g7] [1srb2g]
    • Cloud ecosystem surface: Amazon EC2 DL1 instance family page featuring “Gaudi accelerators from Habana Labs (an Intel company)”. [qnw5d4] [rpzmf3]
    • Company/funding profile: StartupHub.ai profile for “Habana Labs — $75M Raised”. [nof3g7] [1srb2g]

Mission and Identity

  • Habana Labs is consistently described as serving enterprises and cloud providers that need faster and more cost-effective AI compute for training, fine-tuning, and inference of generative AI and other deep learning models. [1srb2g] [qnw5d4] It positions itself around unlocking “order of magnitude improvements in processing performance, cost and power consumption” for AI workloads by designing processors from the ground up for training and inference. [61jiu6] [1srb2g] Its identity today is tightly integrated with Intel’s data-center AI strategy while retaining the Habana Gaudi brand as a specialized accelerator line. [db0lbb] [2rsn74]
  • Stated values / principles:
    • External descriptions emphasize performance, efficiency, and scalability for AI workloads, with messaging around “high-performance, high-efficiency AI accelerators” and optimization “for the specific needs of training deep neural networks and for inference deployment in production environments,” but there is no explicit standalone values list published by the company. [61jiu6] [1srb2g]

What They Do

Habana Labs designs and delivers AI accelerator hardware and software platforms for data centers and cloud providers, focusing on training and inference of deep neural networks and generative AI models. [nof3g7] [1srb2g] It generates value by selling Gaudi-based accelerator cards and systems (now branded Intel Gaudi) plus the SynapseAI software stack that enables developers and infrastructure teams to train and deploy AI models more efficiently than on general-purpose GPUs in some workloads, often via cloud instances like AWS EC2 DL1. [g85rft] [qnw5d4] [rtnhj5]
  • Intel Gaudi training accelerators: Second-generation Gaudi2 and related Gaudi family devices are deep learning accelerators “developed by Habana Labs (now part of Intel)” and “designed from the ground up for large-scale AI training,” featuring memory-centric architecture and high-bandwidth Ethernet-based scale-out. [g85rft] [2rsn74]
  • Goya inference accelerators: Goya processors are described as “optimized for inference,” providing specialized hardware for production deployment of deep learning models in data centers. [rtnhj5] [lrbfe5]
  • SynapseAI software stack: SynapseAI is described as Habana’s “comprehensive software ecosystem for Gaudi processors, providing everything needed to program, optimize, and run machine learning workloads efficiently.” [g85rft]
  • Cloud instances (AWS EC2 DL1): Amazon EC2 DL1 instances are “powered by Gaudi accelerators from Habana Labs (an Intel company)” and are marketed as delivering low cost-to-train deep learning models for NLP, object detection, and image recognition. [qnw5d4] [rpzmf3]
  • On-prem and HPC integrations: Gaudi2 nodes are deployed in HPC environments (e.g., a Gaudi2 node for researchers on a Future Technology Platform), where users can evaluate distributed AI training on Habana Processing Units (HPUs). [g85rft]
  • AI infrastructure for science and engineering: Academic platforms like SDSC’s Voyager system “based on Intel’s Habana Lab AI processors” use Gaudi (training) and Goya (inference) processors with high-speed networking to explore AI in science and engineering. [rtnhj5]
  • Enterprise and cloud AI solutions: StartupHub.ai characterizes Habana as operating in computing, artificial intelligence, data center, training, inference, and AI infrastructure segments, serving enterprises and cloud providers needing scalable AI compute. [1srb2g] [nof3g7]

Leadership and People

  • Avigdor Willenz — founder and prominent Israeli semiconductor entrepreneur; multiple sources attribute the founding team of Habana Labs to him, and later note he leads another chip startup “under the founding team of Habana Labs,” indicating his key role in its origin. [6kuyzb] [7vx0xk]
  • Intel corporation — parent company; Intel acquired Habana Labs in 2019 as part of its broader AI accelerator strategy, integrating Habana into Intel’s Data Products Group. [7vx0xk] [db0lbb]
  • Named current CEO or specific C‑suite titles for Habana Labs as a distinct entity are not clearly disclosed in the searched results; Intel generally presents Habana as an internal AI accelerators business unit rather than a separately branded leadership structure. [db0lbb] [1srb2g]

History and Origin Story

Habana Labs was founded in 2016 in Israel to build AI and deep learning processors for data centers and cloud environments, targeting more efficient training and inference than general-purpose hardware. [1srb2g] [lrbfe5] The company launched its first inference and training products in 2018 and 2019, respectively, and was acquired by Intel in 2019 (deal closing reported in early 2020 in some summaries), after which it has operated as an autonomous Intel subsidiary focused on the Gaudi and Goya accelerator lines. [7vx0xk] [on61d7] [db0lbb]
  • 2016 — Habana Labs founded in Israel to develop disruptive AI and deep learning computing solutions for data center cloud and on-premise. [lrbfe5] [1srb2g]
  • 2018 — Launch of Habana products for AI inference, marking its first commercial hardware for production deployment. [lrbfe5]
  • 2019 — Launch of Habana products for AI training, expanding from inference into full deep learning training accelerators. [lrbfe5]
  • December 2019 — Intel announces acquisition of Habana Labs; coverage notes “In December 2019 Intel bought Habana Labs,” and Intel later positions the deal as central to its AI accelerator portfolio. [on61d7] [db0lbb]
  • 2020 — Job and profile descriptions cite that Habana “was acquired by Intel in 2020,” and began operating as an autonomous subsidiary within Intel’s Data Products Group. [lrbfe5] [7vx0xk]
  • 2020s — Gaudi-based accelerators gain adoption in cloud and HPC, e.g., powering AWS EC2 DL1 instances and academic systems like Voyager for AI training and inference at scale. [qnw5d4] [rtnhj5]

Financials and Funding

As a private company acquired by Intel, Habana Labs does not report standalone public financials, but its venture funding prior to acquisition is documented.
Funding rounds (pre-acquisition)StartupHub.ai and other profiles indicate Habana raised significant venture funding before Intel’s purchase; however, detailed round-by-round terms are not fully enumerated in high-quality primary sources, so only aggregate figures are used.
RoundDateAmountLead investor
Series ANot reliably specifiedNot reliably specifiedNot reliably specified
Series BNot reliably specifiedNot reliably specifiedNot reliably specified
Total—Reported total funding approximately $75M–$120M across rounds [nof3g7] [1srb2g] [xv79zu] —
  • Investors (alphabetical, where cited): External profiles list aggregate total funding of $75M in some startup rankings and $120M in StartupHub.ai’s company profile; specific investor names are not consistently disclosed in the search results used here, so no reliable investor list can be provided. [nof3g7] [61jiu6] [1srb2g] [xv79zu]

Milestones and Signature Output

  • Gaudi (first-generation training accelerator) — ~2019 — Initial Gaudi processors introduced as deep learning accelerators designed from the ground up for large-scale AI training, marking Habana’s entry into the training hardware market. [g85rft] [2rsn74] [lrbfe5]
  • Goya inference accelerator — ~2018–2019 — Goya processors launched as AI inference accelerators optimized for production deployment, distinguishing Habana’s product line between training and inference. [rtnhj5] [lrbfe5]
  • Intel acquisition — 2019–2020 — Intel’s acquisition of Habana Labs integrated its Gaudi and Goya accelerators into Intel’s data-center AI portfolio and led Intel to discontinue its prior Nervana NNP‑T line in favor of Habana technology. [on61d7] [db0lbb]
  • Gaudi2 launch — early–mid 2020s — Gaudi2 described as Intel’s second-generation deep learning accelerator “developed by Habana Labs (now part of Intel),” with 96 GB of high-bandwidth memory and 24×100 Gbps Ethernet interfaces, enabling efficient training of large and complex models. [g85rft] [2rsn74]
  • AWS EC2 DL1 general availability — 2020s — Amazon introduces EC2 DL1 instances “powered by Gaudi accelerators from Habana Labs (an Intel company),” providing a cloud-based, low cost-to-train option for deep learning workloads. [qnw5d4] [rpzmf3]
  • Adoption in HPC / scientific AI — 2020s — Systems like SDSC’s Voyager and GWDG’s Future Technology Platform integrate Habana Gaudi and Goya processors to support AI in science and engineering, with high-speed networking and scale-out capabilities. [rtnhj5] [g85rft]

Ecosystem and Relationships

  • Intel — Parent company; Intel acquired Habana Labs in 2019 and integrated it into Intel’s Data Products Group and overall AI accelerator strategy. [on61d7] [db0lbb] [7vx0xk]
  • Amazon Web Services (AWS) — Cloud partner; AWS EC2 DL1 instances are explicitly “powered by Gaudi accelerators from Habana Labs (an Intel company).” [qnw5d4] [rpzmf3]
  • HPC centers and research institutions — Partners; environments like GWDG’s Future Technology Platform and SDSC’s Voyager cluster deploy Gaudi and Goya processors for AI research and education. [g85rft] [rtnhj5]
  • Competing AI chip vendors — Ecosystem peers/competitors include companies like Qualcomm and other data-center AI chip makers referenced in industry analyses of AI accelerators versus GPUs, though specific direct rival lists are not enumerated in the cited results. [dpi5pv] [2rsn74]
  • Element Labs — A newer Israeli chip startup described as “operates in complete secrecy under the founding team of Habana Labs,” highlighting that Habana’s founding talent has gone on to seed additional ventures in the AI chip ecosystem. [6kuyzb]

Recent Developments

As of 2026-08-06,
  • 2026‑07‑17 — Industry commentary on AI chips notes that “In December 2019 Intel bought Habana Labs” and emphasizes Intel’s pivot away from Nervana to Habana technology as context for current AI hardware strategy and consolidation. [on61d7]
  • 2026‑07‑30 — Coverage of Intel-led investment in SambaNova Systems references Intel’s prior AI accelerator moves, including “the 2019 acquisition of Habana Labs,” positioning the deal as part of Intel’s ongoing portfolio of AI hardware bets. [db0lbb]
  • 2026‑07‑09 — A neuromorphic computing patent analysis notes Intel’s AI accelerator acquisitions and states that Intel “acquired Habana Labs for AI accelerators,” distinguishing Habana’s role from Intel’s neuromorphic efforts like Hala Point. [fth876]
  • 2026‑07‑09 — Academic and HPC platforms continue to highlight systems “based on Intel’s Habana Lab AI processors” (Gaudi for training and Goya for inference), indicating ongoing use of Habana hardware in AI for science and engineering. [rtnhj5]
  • 2026‑07‑24 — Job description text for positions in Israel reiterates that Habana “was acquired by Intel in 2020” and operates as an autonomous subsidiary within Intel’s Data Products Group, suggesting continued hiring and product development. [lrbfe5] [7vx0xk]

Impact

  • Impact on society:
    • By enabling lower-cost AI training via cloud instances like AWS EC2 DL1, powered by Habana’s Gaudi accelerators, Habana contributes to broader access to deep learning for organizations that might otherwise be constrained by GPU costs, particularly in areas like NLP, object detection, and image recognition. [qnw5d4] [rpzmf3]
    • Habana-backed systems like SDSC’s Voyager and GWDG’s Gaudi2 node provide researchers and students with access to AI-focused hardware for science and engineering applications, impacting academic and scientific communities. [rtnhj5] [g85rft]
  • Impact on innovation:
    • Habana Labs helped pioneer Ethernet-connected, memory-centric AI training accelerators (Gaudi and Gaudi2) with standard 100 Gbps Ethernet scale-out, offering an alternative architecture to traditional GPU-based solutions for large-scale AI training. [g85rft] [2rsn74] [rtnhj5]
    • Intel’s decision to discontinue its Nervana NNP‑T in favor of Habana technology underscores Habana’s influence on Intel’s AI hardware roadmap and, by extension, on the broader AI accelerator market. [on61d7] [fth876]
  • Impact on its industry or domain:
    • Habana’s Gaudi accelerators, via offerings like EC2 DL1, have become one of the main non‑GPU options for large-scale AI training, pushing competitors and cloud providers to consider heterogeneous accelerator portfolios beyond dominant GPU vendors. [qnw5d4] [2rsn74] [dpi5pv]
    • Industry and investment analyses frequently cite the acquisition of Habana Labs as a landmark AI chip deal, using it as a benchmark for later valuations and strategic moves in the AI accelerator sector. [xv79zu] [db0lbb]
  • Historical significance:
    • The 2019 Intel–Habana deal is routinely referenced as a key moment in the AI chip “arms race,” illustrating the shift of major CPU vendors into specialized AI accelerators and shaping how subsequent AI hardware startups and acquisitions are evaluated. [on61d7] [db0lbb] [xv79zu]
  • Criticisms and controversies:
    • No substantive public controversies or regulatory findings specifically targeting Habana Labs were identified in the search results; most external commentary treats the company positively as a strategic AI hardware acquisition by Intel. [on61d7] [fth876] [db0lbb]

Adjacent Entries

  • Intel — Parent company and integrator of Habana’s Gaudi and Goya accelerators into its data-center AI portfolio. [on61d7] [db0lbb]
  • Amazon Web Services — Cloud provider offering EC2 DL1 instances powered by Habana Gaudi accelerators. [qnw5d4] [rpzmf3]
  • NVIDIA — NVIDIA — Dominant GPU vendor and primary competitive benchmark for Habana’s AI accelerators in data-center training and inference. [dpi5pv] [2rsn74]
  • Qualcomm — Qualcomm — Another AI chip rival mentioned in industry analyses of data-center AI hardware competition. [dpi5pv]
  • SambaNova Systems — AI accelerator startup whose Intel-led funding round is often contextualized alongside Intel’s prior acquisition of Habana Labs. [db0lbb]
  • Element Labs — Israeli AI chip startup associated with the founding team of Habana Labs, illustrating the continuing entrepreneurial impact of Habana’s founders in the semiconductor ecosystem. [6kuyzb]

Sources