Full-Time
Devices Software Engineer
Embedded
Posted on 8/5/2024
Groq
201-500 employees
High-speed AI inference hardware platform
Compensation Overview
$208.8k - $365.4k/yr
Mountain View, CA, USA
Remote
Some roles may require being located near or on our primary sites, as indicated in the job description.
 Software Engineering (1) |
|---|
- Minimum B.S. Degree in Computer Science or Electrical Engineering or Computer Engineering, or equivalent practical experience.
- Over five years programming low level programming skills in C/C++.
- 5 years experience developing embedded systems.
- Proficiency with debugging large, complex systems and hardware/software interfaces.
- Experience with communication protocols: I2C, CANBUS, SPI, USB.
- Design and develop software/firmware for a variety of architectures (x86, ARM, 8051, RiscV).
- Build and deploy infrastructure to debug and validate functionality at the individual component and system level.
- Design and implement messaging protocols that allow various components to communicate on a system and board level.
- New hardware/product bring-up.
- Linux BSP bring-up and drivers
- Buildroot and makefile build systems
- Python
- FPGAs and high speed data interconnects
Groq builds AI inference hardware and systems, led by the Groq LPU, to run machine learning models with high throughput and low energy use. It offers scalable AI inference for both cloud and on-premises deployments, with design, fabrication, and assembly done in North America to ensure quality. The Groq LPU is a hardware accelerator that delivers fast, energy-efficient model execution, supported by integrated systems engineering and services. Its focus on domestic manufacturing and proven performance per watt helps enterprises run large-scale AI workloads quickly and cost-effectively across data centers and edge-like environments.
Company Size
201-500
Company Stage
Acquired
Total Funding
$23.3B
Headquarters
Mountain View, California
Founded
2016
Simplify's Take
What believers are saying
- Demand for agentic AI favors faster, cheaper inference at scale.
- Samsung Foundry’s mature 4nm production supports Groq’s next LPU shipments.
- GroqCloud can monetize developers before hardware deployments fully ramp.
What critics are saying
- Nvidia’s licensing deal embeds Groq’s IP inside a larger ecosystem, weakening standalone differentiation.
- Samsung Foundry concentration creates a single-point supply risk for every major LPU launch.
- Groq lacks training capability, leaving customers to standardize on Nvidia or Google for full-stack AI.
What makes Groq unique
- Groq’s LPU targets inference, not training, for deterministic low-latency token generation.
- The company pairs custom silicon with GroqCloud, offering public, private, and co-cloud access.
- Its chips emphasize low power, on-chip SRAM, and simple scalable data-center integration.
Help us improve and share your feedback! Did you find this helpful?
Your Connections
People at Groq who can refer or advise you
Benefits
Remote Work Options
Company Equity
Growth & Insights and Company News
6 month growth
↓ -1%1 year growth
↑ 15%2 year growth
↑ 14%Read weekly Deep Dives from my research team and join my Learn With Me Substack community. I'm looking to share my knowledge base and AI agents that I use over there soon: https://chamath.substack.com/ Legal Disclaimer: Any information on this channel is provided for educational purposes only. None of the above is investment advice, a recommendation or an offer to sell, or a solicitation of an offer to buy, any securities or investment products.
Samsung-backed AI chip firm Rebellions raises $400 million ahead of IPO. Published Mon, Mar 30 20269:00 AM EDT Key Points * South Korean AI chip startup Rebellions raised $400 million at a valuation of $2.34 billion. * The company plans to use the funds to expand in the U.S. as it prepares for an initial public offering. * Rebellions' chips are focused on AI inferencing where it competes with Nvidia as well as other startups from Groq to Cerebras. The Rebel-Quad is the second-generation product from Rebellions and is made up of four Rebel AI chips. Rebellions, a South Korean firm, is looking to rival companies like Nvidia in AI chips. Rebellions South Korean AI chip startup Rebellions said Monday it has raised $400 million as it looks to expand into the U.S. market ahead of a public listing. Mirae Asset Financial Group and the Korea National Growth Fund, an investment vehicle of the South Korean government, led the round, which values Rebellions at $2.34 billion. Rebellions is one of the many semiconductor startups looking to capitalize on demand for AI chips and investor appetite for companies that are fueling the build-out of infrastructure for the technology. Sunghyun Park, CEO of Rebellions, told CNBC that the money will be used to expand into the U.S. "Our main target right now is big labs," Park said, naming companies like Meta and xAI as target customers, rather than hyperscalers like Amazon and Microsoft. Park added that Rebellions currently has some active proof-of-concept trials with customers in the U.S. The CEO also said the company is preparing for an initial public offering, as CNBC previously reported, but declined to give any specifics on the timeline or listing location. AI inference focus. Rebellions' chips are focused on inferencing, the process of running AI applications rather than training them. While Nvidia's graphics processing units (GPUs) have been the gold standard for training AI models, there is an increasing focus on chips that can run inferencing processes quickly while also being more energy efficient. Rebellions sells server systems made up of its Rebel100 NPU chips. The South Korean startup competes not only with Nvidia but also with a growing list of other startups from Cerebras to Groq - a company that Nvidia licenses technology from. "When it comes to inference alone, our chip offers... much higher energy efficiency and performance at the same time," Park said, discussing Rebellions' differentiation with the competition. Park declined to provide sales figures but said that the company has a "strong revenue pipeline." However, he noted that one of the challenges right now is securing a supply of memory chips. These types of semiconductors, produced by Samsung, SK Hynix and Micron, are in high demand but also in short supply, leading to an unprecedented rise in prices of the component. "Memory is not very easy to get. But our demand is so huge," Park said. He added that because two of the world's biggest memory makers, Samsung and SK Hynix, are investors, Rebellions is "the best-positioned" to obtain memory supply, compared to other startups. South Korea's chip bet. Rebellions is a central part of the South Korean government's attempt to boost the country's domestic semiconductor sector. Last year, the government launched its "K-Nvidia" initiative, a plan designed to invest government funds into companies designing advanced AI chips. The Korea National Growth Fund, one of Rebellions' investors in the latest funding round, contributed 250 billion Korean won ($166 million), the government announced on Friday. Samsung, SK Hynix and Saudi Arabian oil giant Aramco are all investors in Rebellions.
Nvidia revealed a major partnership with Groq at its GTC 2026 conference, integrating Groq's low-latency inference technology into its AI stack. Following a multibillion-dollar licensing deal, the Groq 3 inference accelerator will serve as a complementary technology to Nvidia's GPUs rather than a competing architecture. The announcement marks a strategic shift for Nvidia, which has dominated AI hardware through its Blackwell and Hopper chip architectures over the past five years. The company's GPUs have powered most large language model training, whilst its broader ecosystem includes CUDA accelerators and NVLink technology. The collaboration positions Nvidia to maintain its central role in the AI boom, as tech giants race to build advanced data centres and AI startups continue emerging across various markets.
The best stocks to invest $5,000 in right now. Geoffrey Seiler, The Motley Fool If you've got $5,000 available to invest that isn't needed to bolster an emergency fund or pay down short-term debt, and are looking to get into the market, I would start by investing in two leading artificial intelligence (AI) stocks. Let's look at two great options you can buy right now and hold for the long term. $3,074,639 Volume Ends on Jun 30, 2026 Loading chart... Nvidia. The king of AI infrastructure, Nvidia (NASDAQ: NVDA), is not sitting still waiting for its crown to be usurped. The company has grown to be the largest in the world on the back of its graphics processing units (GPUs), and it is still seeing rapid revenue growth. Its CUDA software platform has been a differentiator and created a wide moat, but it knows that this may no longer be enough by itself as the market shifts more toward inference and agentic AI. Will AI create the world's first trillionaire? Its team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue" Through its acquisition of Groq and SchedMd, Nvidia has really positioned itself to be a leader in these two fields and expand on its ecosystem advantages. Penneco Pipeline Corp. can already see this with its introduction of NemoClaw, which utilizes SchedMD's Slurm, to create a powerful agentic AI platform, and by integrating Groq's inference-focused LPUs (language processing units) into its chip and software platform. Nvidia is no longer just a chip company; it's become an end-to-end AI data center behemoth with one of the most complete tech stacks out there. That makes it a solid long-term investment. Alphabet. In the AI revolution, it's becoming more and more about controlling the ecosystem. Nvidia is doing that on one end, while Alphabet (NASDAQ: GOOGL) (NASDAQ: GOOG) is attacking it on the other end. Alphabet has the most complete AI stack at the moment, as it is the only company with world-class AI chips, with its tensor processing units (TPUs) plus a top-tier AI model in Gemini. Alphabet's biggest advantage is its TPUs, which were developed more than a decade ago and which it has long used to power much of its internal workloads. These chips help Alphabet be one of the few companies not beholden to Nvidia, which ultimately gives it a huge cost advantage. By using its own chips, it can train its models and run inference much more cost-effectively than competitors who primarily use GPUs. Meanwhile, that advantage just increases as spending on AI infrastructure rises. This creates a powerful flywheel effect that positions Alphabet to be a long-term AI leader.
Nvidia has revealed why it spent $20 billion to license AI chip startup Groq's technology and hire its engineers: speed to market. The company's newly announced Groq 3 LPX racks, packing 256 LP30 language processing units into a single system, are based on Groq's second-generation LPU technology with minimal modifications. The LP30 chip delivers 1.2 petaFLOPS of FP8 compute and 500MB of on-chip SRAM memory achieving speeds up to 150TB/s. However, it lacks Nvidia's NVLink interconnect, NVFP4 hardware support, and CUDA compatibility at launch. LPX racks are designed to pair with Nvidia's Vera-Rubin NVL72 systems, splitting inference workloads between GPUs and LPUs. The platform targets hyperscalers and model developers serving trillion-parameter models at token rates exceeding 500 to 1,000 per second, requiring between four and eight LPX racks depending on precision.