xAI brought its Colossus 2 AI supercomputer online in January 2026 and said the combined Colossus facilities had become the first coherent AI-training cluster to reach approximately one gigawatt of compute power. The system supports training and inference for Grok, but the “world-first” distinction remains an xAI and Elon Musk claim rather than an independently audited global ranking.
What xAI launched
Musk announced on January 17, 2026, that Colossus 2 was operational and described it as the “first Gigawatt training cluster in the world.” A January 19 report said the system had crossed the one-gigawatt threshold and could expand further. The reported milestone concerns a large, interconnected AI-training system rather than a single conventional computer.
The naming can be confusing:
- Colossus 1 is xAI’s original Memphis, Tennessee, cluster.
- Colossus 2 refers to the subsequent expansion associated with the January 2026 milestone.
- The combined Colossus and Colossus II platform includes multiple facilities, hardware generations and power systems. A later company filing describes the combined sites as one coherent gigawatt-scale training cluster.
The broader footprint includes South Memphis and facilities in Southaven, Mississippi. The original Memphis site occupies a converted former factory of roughly one million square feet. xAI says the location helped it move quickly because of existing industrial infrastructure, available space and local utility relationships. xAI’s Memphis overview provides the company’s account of the buildout.
What “one gigawatt” means
A watt measures instantaneous power. One gigawatt equals one billion watts. In an AI data-center story, however, “one gigawatt” does not automatically mean the facility continuously consumes one billion watts from the public grid.
#1 Best Overall
- EVOLUTION CORE ULTRA 9 285H MINI PC - GMKtec EVO-T1 is the next evolution in AI mini PC Ultra 9 series. The Core Ultra 9 285H offers 16 cores (six P-cores + eight E-cores + two LPE-cores) and 16 threads with a turbo clock of 5.4 GHz. It is currently one of the best value for performance AI mini PC computers.
- AI NPU - The 285H features an Intel AI Boost NPU, capable of up to 13 TOPS (Tera Operations per Second) for INT8 calculations, which is designed to accelerate AI tasks.
- INTEL ARC 140T GAMING PC - The Arc 140T GPU includes 8 Xe cores and supports features like DirectX 12, OpenGL 4.5, and OpenCL 3, making it capable of handling modern games and creative applications. It also supports Quick Sync Video for efficient video encoding and decoding, as well as AV1 encoding and decoding.
- 64GB DDR5 RAM + 1TB SSD - The EVO-T1 is equipped with Dual 32GB (Total 64GB) SO-DIMM DDR5 5600MHz memory sticks. 2TB PCIE 4.0 SSD Drive with 3x M.2 2280 Expansion slots. Each slot capable of reading up to 4TB. (12TB MAX)
- QUAD SCREEN 8K DISPLAY SUPPORT - EVO-T1 AI Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and USB Type-C Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support.
The figure might refer to:
- Power delivered to GPUs and other compute hardware;
- Total electricity used by the facility, including cooling and networking;
- Contracted or available electrical capacity;
- Planned future capacity; or
- A combination of sites treated as one training fabric.
A later first-party filing says Colossus and Colossus II collectively provide approximately 1.0 gigawatt of compute power, with additional capacity for facility operations. That is the clearest formulation available, and it should not be rewritten as proof that Colossus continuously draws a full gigawatt of electricity. The filing also describes the sites, rather than one building, as the coherent cluster.
A comparison with the electricity demand of a large city can illustrate the scale, but it is only a rough analogy. A data center and a city have different load profiles, infrastructure and operating patterns.
The hardware numbers are changing—and inconsistent
xAI’s published GPU figures vary according to the date, facility, accelerator generation and whether the company is describing physical processors or broader capacity. The numbers should not be added together as though they were one definitive inventory.
| Source or phase | Reported figure | How to read it |
|---|---|---|
| Original Colossus announcement | 100,000 NVIDIA Hopper GPUs, with an expansion planned to 200,000 | Earlier buildout figures |
| Current Colossus page | 200,000 interconnected H100 GPUs; another section says 180,000 | Internal inconsistency on xAI’s own page |
| Later mixed-GPU announcement | More than 220,000 GPUs, including H100, H200 and GB200 accelerators | A later mixed-generation configuration |
| Colossus II filing | About 110,000 GB200 processors and about 110,000 GB300 processors across two clusters | Separate Colossus II configurations |
The practical conclusion is that xAI’s infrastructure has evolved from an H100-heavy system into a mixed-generation cluster using H100, H200, GB200 and GB300-class NVIDIA accelerators. The exact total depends on which sites and expansion phase are being counted. Relevant company announcements are available from the original Colossus announcement, the current Colossus page and the later mixed-GPU announcement.
Recommended Free Tools
Why Grok needs a cluster this large
Frontier AI development requires more than a single training run. Large clusters can be used for:
- Pretraining models on large datasets;
- Reinforcement learning and reasoning improvements;
- Fine-tuning and evaluation;
- Running many experiments in parallel; and
- Serving inference requests after a model is deployed.
In distributed training, many GPUs work on the same model. A tightly interconnected cluster reduces the communication delays involved in exchanging data between those processors. That can let researchers run larger experiments, complete training runs faster or perform more experiments within the same period.
Rank #2
- LOW ENERGY HIGH PERFORMANCE MINI PC - The Intel Core Ultra 5 125U is part of the Ultra 5 lineup, using the Meteor Lake architecture with BGA 2049. Intel Hyper-Threading technology is available and effectly doubles the core-count of the P-Cores, to a total of 14 threads. Core Ultra 5 125U has 12 MB of L3 cache and operates at 1300 MHz by default, but can boost up to 4.3 GHz, depending on the workload. With a TDP of 15 W, the Core Ultra 5 125U consumes very little energy but outputs high performance efficiency
- 32GB DDR5 RAM + 512GB SSD - The K15 mini computer is equipped with Dual 16GB (Total 32GB) SO-DIMM DDR5 4800MHz memory sticks. 512GB PCIE 4.0 SSD Drive with 3x M.2 2280 Expansion slots. Each slot capable of reading up to 8TB. (24TB MAX)
- QUAD SCREEN 4K DISPLAY SUPPORT - K15 Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and USB Type-C Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support
- OCULINK PORT - The Oculink port on the rear interface enables higher bandwidth capabilities, better frame rates and lower lag. The standard also operates at PCIe x4 speeds, compared to Thunderbolt's x3. Gamers and content creators can benefit from Oculink's higher bandwidth, resulting in better performance and lower lag for eGPU setups
- DUAL NIC FAST 2.5GBE + WIFI 6E + BT 5.2 - Dual Ethernet 2.5GbE LAN port design provides more applications, such as firewall, multichannel aggregation, soft routing, file storage server. Built-in WIFI 6E / Bluetooth 5.2 is more stable and efficient to connect multiple wireless devices such as projector, printer, monitor, speakers and etc
xAI says its Grok 4 models were trained on Colossus and that it used large-scale reinforcement-learning compute to improve reasoning and agency. The company’s financing announcement places Colossus in the broader strategy of building infrastructure for Grok.
More hardware does not automatically mean a more capable model. Results also depend on training data, algorithms, software efficiency, model design, safety systems and product engineering. The Colossus milestone demonstrates industrial scale; it does not, by itself, prove a particular improvement in Grok’s accuracy or reliability.
How quickly was it built?
xAI has emphasized speed as a competitive advantage. The company says the original Colossus was built in approximately 122 days, or four months, and began running workloads 19 days after the first servers arrived. Its current Colossus page says the first expansion reached approximately 200,000 GPUs in another 92 days.
The later filing gives different but related figures for Colossus II: about 91 days for its first cluster and 64 days for a second cluster. These are company-reported timelines, not independently audited construction benchmarks.
Rapid deployment requires more than installing servers. The project also needs electrical generation and distribution, cooling, batteries, networking, fiber connections, maintenance systems, permits and utility coordination. Reusing an industrial building can shorten the construction schedule, but it does not remove those engineering and regulatory challenges.
How the facilities are powered
xAI has used a combination of behind-the-meter generation, grid connections, batteries and utility partnerships. Its Memphis materials describe 35 natural-gas turbines, a possible future expansion to as many as 90 turbines, two planned grid substations and more than 240 batteries at the Electrolux site. Some generators are described as backup or supplemental sources when the grid is strained. xAI’s leadership page outlines the company’s power and water plans.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsRank #3
- Entry-level NAS Personal Storage：UGREEN NAS DH2300 is your first and best NAS made easy. It is designed for beginners who want a simple, private way to store videos, photos and personal files, which is intuitive for users moving from cloud storage or external drives and move away from scattered date across devices. This entry-level NAS 2-bay perfect for personal entertainment, photo storage, and easy data backup (doesn't support Docker or virtual machines).
- Set Your Devices Free, Expand Your Digital World: This unified storage hub supports massive capacity up to 64TB.*Storage drives not included. Stop Deleting, Start Storing. You can store 22 million 3MB images, or 2 million 30MB songs, or 43K 1.5GB movies or 67 million 1MB documents! UGREEN NAS is a better way to free up storage across all your devices such as phones, computers, tablets and also does automatic backups across devices regardless of the operating system—Window, iOS, Android or macOS.
- The Smarter Long-term Way to Store: Unlike cloud storage with recurring monthly fees, a UGREEN NAS enclosure requires only a one-time purchase for long-term use. For example, you only need to pay $459.98 for a NAS, while for cloud storage, you need to pay $719.88 per year, $2,159.64 for 3 years, $3,599.40 for 5 years. You will save $6,738.82 over 10 years with UGREEN NAS! *NAS cost based on DH2300 + 12TB HDD; cloud cost based on 12TB plan (e.g. $59.99/month).
- Blazing Speed, Minimal Power: Equipped with a high-performance processor, 1GbE port, and 4GB RAM on Board, this NAS handles multiple tasks with ease. File transfers reach up to 125MB/s—a 1GB file takes only 8 seconds. Don't let slow clouds hold you back; they often need over 100 seconds for the same task. The difference is clear.
- Let AI Better Organize Your Memories: UGREEN NAS uses AI to tag faces, locations, texts, and objects—so you can effortlessly find any photo by searching for who or what's in it in seconds. It also automatically finds and deletes similar or duplicate photo, backs up live photos and allows you to share them with your friends or family with just one tap. Everything stays effortlessly organized, powered by intelligent tagging and recognition.
This approach highlights the trade-off behind rapid AI infrastructure expansion. On-site generation can reduce dependence on a delayed grid interconnection and provide power sooner. Natural-gas turbines also bring emissions, noise, permitting and local-air-quality concerns. Batteries and grid connections can improve flexibility, but they do not eliminate the need for substantial generation and transmission infrastructure.
Environmental and community questions
The Memphis project raises several questions that are separate from whether the cluster is technically impressive:
- Air quality: Natural-gas generation produces emissions, and xAI’s own fact-versus-fiction page acknowledges that Colossus will be among Memphis’s largest emissions sources.
- Water: AI data centers may require substantial cooling resources. xAI says it is developing an $80 million water-recycling plant and projects that it could save 4.745 billion gallons annually, or reduce strain on the Memphis Aquifer by a claimed 9 percent.
- Noise and land use: Turbines, cooling equipment, construction and industrial operations can affect nearby communities.
- Electricity infrastructure: New substations, transmission work and utility upgrades raise questions about who pays and whether costs are passed to ratepayers.
- Local benefits: Construction, operations and subcontracting can create jobs, but the scale and durability of those benefits should be evaluated separately from the company’s investment claims.
The water-saving figure is a company projection, not an independently verified outcome. Similarly, xAI’s claims about grid effects, emissions controls and infrastructure costs should be distinguished from measurements by regulators or independent analysts. The company’s response to public criticism sets out its position but does not settle every community or environmental question.
Is it really the world’s first?
The safest answer is: xAI says it is.
The “world-first” claim originated with Musk and xAI and was repeated in secondary reporting. The available sources do not provide an independent audit comparing every major AI cluster worldwide. The answer can also change depending on what is being compared:
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →- A single building or several connected facilities;
- GPU compute power or total facility electrical capacity;
- Operational capacity or planned capacity;
- A tightly interconnected training fabric or an organization’s total global compute; or
- Physical GPU count or equivalent performance.
For that reason, “xAI’s first gigawatt-scale AI training cluster” or “xAI says Colossus 2 was the world’s first operational gigawatt-scale AI training cluster” is more accurate than presenting the superlative as independently settled fact. The later filing’s narrower phrase—“coherent gigawatt-scale AI training cluster”—better describes the claim’s scope.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What comes next
The filing describes additional planned capacity, including at least 220,000 more GB300 processors and more than 400 megawatts of power. Future targets should not be confused with deployed capacity. Likewise, claims about 1.5 gigawatts, two gigawatts or one million GPUs should be labeled as plans or projections unless a source confirms that they are operational.
Rank #4
- [Powerful PC] Gaming PC equipped with Core i9-14900F, 24 Cores 32 Threads, 36M Cache, Max Turbo Frequency: 5.8GHz, Windows 11 pro (64 Bit). With GeForce RTX 50 Series GPUs. Adopting DLSS 4 technology, it dramatically improves frame rate performance, supports FP4 low-precision computing, and doubles the efficiency of AI inference. SD graph generation speed is 3 times faster than RTX 4070 Super, significantly increasing creative productivity. Graphics work productivity has increased significantly.
- [High Speed DDR5 RAM & PCIE4.0 SSD] The desktop computer is equipped with Dual-DDR5 RAM (dual channel DDR5 high-speed memory, which can support up to 128GB RAM), 1 x M.2 2280 PCIE4.0 high-speed SSD, and support add 2 x 2.5-inch SATA HDD/SSD(not include) is enough to accommodate system files and massive games, Excellent reading and writing speed greatly shortening your boot time.
- [8K@60Hz Quad-Display] Desktop PC with GeForce RTX 5070 12G GDDR7, supporting DLSS 4, ray tracing, and AI cores. Easily connect 4 monitors via 1×HDMI 2.1 + 3×DP 1.4a — all ports support 8K@60Hz. Delivers stunning visuals and ultra-smooth performance for home entertainment, live streaming, video editing, AI workloads, 3D rendering, and AAA gaming.
- [Functional Interfaces] Mini computer is equipped with 4 x USB 3.2, 4 x USB2.0, 1 x HDMI2.1 port, 3 x DP ports, 2xRJ-45 Gigabit Network Ethernet, 1 x Fiber Optic PORT, 1 x Audio in/out. Built-in Bluetooth 5.4 and IEEE 802.11be wifi 7, Higher transfer rates and lower latency. Mini PC supports multiple device connection and can be used with servers, monitoring equipment, office equipment, projectors, televisions, etc, Mini desktop computer support automatic power on and Wake On Lan.
- [Warranty & Liquid Cooling] Warrant: 2 year/24 months. The compact computer size: 11.6*9.3*3.9in, 9.25lb, Chassis built-in 2 large copper fans, built-in liquid cooling device, to further enhance the computer heat dissipation, and at the same time can reduce noise, give full play to the overall performance of the computer.
xAI is also positioning Colossus as more than an internal research facility. In May 2026, the company announced an agreement to provide Anthropic with access to Colossus 1 compute. That establishes a commercial compute-access arrangement, not a public self-service marketplace or publicly posted rental price. The announcement is available from xAI.
What the milestone means for businesses and investors
Colossus illustrates the capital intensity of frontier AI. Companies pursuing this strategy need access to accelerators, high-speed networking, power generation, cooling, industrial sites, specialized staff and long-term financing. Controlling more of that stack can reduce dependence on cloud providers and improve scheduling for large training runs.
Free tools Windows power users keep installed
One-click scans. No signup required.
The trade-off is flexibility and cost. A dedicated supercluster can be highly efficient for large, sustained training workloads but less suitable for small jobs, geographically distributed inference or customers that need usage-based billing. Renting cloud GPUs offers faster initial access and geographic redundancy, while ownership can offer greater control but requires much larger upfront investment and operational expertise.
For personal-finance readers, the relevant lesson is not that consumers can buy a share of Colossus hardware. It is that AI services increasingly depend on large, power-intensive infrastructure projects whose costs, financing, energy needs and regulatory risks can affect technology companies and their suppliers.
The Bottom Line
Bottom line: Colossus 2 appears to have reached the approximately one-gigawatt threshold for a coherent AI-training cluster in January 2026, supporting Grok and future commercial workloads. The achievement is best understood as an infrastructure and industrial-scaling milestone. The precise GPU total, electricity interpretation and “world-first” ranking remain qualified because xAI’s figures vary and no independent global audit has established the superlative.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
