News
CAPACITY TEST:

Intel Lists Starfire Space AI Chip At 75 TOPS As Radiation Data Stays Pending

Newsroom brief

Tom's Hardware reported that Intel has listed Starfire, a space-grade system-on-chip for the U.S. government, with up to 75 TOPS and Intel 18A CPU and NPU tiles. The sell sheet says samples are planned for Q3 2026, while radiation qualification data and final specifications remain pending.

Verified against source materialEdited by SendTech Times Chips & Compute Desk
Intel Lists Starfire Space AI Chip At 75 TOPS As Radiation Data Stays Pending
Image source: tomshardware.com

Intel has listed Starfire as a space-grade AI system-on-chip for the U.S. government with up to 75 TOPS, Tom's Hardware reported, but the same sell sheet still leaves radiation qualification data in progress.

The report said Intel Government Technologies is handling Starfire and that samples are planned for Q3 2026.

The chip combines Intel 18A compute and NPU tiles with an Intel 3 graphics tile in a Foveros package.

Intel Starfire Uses 18A CPU And NPU Tiles

Tom's Hardware reported that Starfire has eight CPU cores and a three-tile NPU on Intel 18A, paired with a four-core Xe GPU built on Intel 3.

Intel lists two versions of the chip: a 10 W low-power version rated at 45 TOPS and a 35 W performance version rated at 75 TOPS.

Tom's Hardware reported that the sell sheet gives the part an operating range from -55 to 125 Celsius, lists 12 PCIe Gen4 lanes, supports LPDDR5 or DDR5 memory, and gives both versions a 10-plus year lifetime rating.

The low-power version runs its P-cores at 1.0 GHz, its efficiency cores at 850 MHz and its GPU between 800 MHz and 1.0 GHz, according to Tom's Hardware.

The performance version raises those clocks to 3.1 GHz for the P-cores, 2.1 GHz for the efficiency cores and 2.0 GHz for the GPU.

Starfire Targets On-Orbit AI Inference Rather Than Telemetry

Tom's Hardware compared Starfire with BAE Systems' RAD750, the radiation-hardened PowerPC processor that has been used for two decades.

Public specifications cited by the outlet list RAD750 at 110 MHz to 200 MHz, 10.4 million transistors and 150nm or 250nm lithography.

The report said RAD750 has flown on the Mars rovers, Kepler, Fermi and more than 150 spacecraft.

Tom's Hardware reported that BAE's multi-core RAD5545 and a NASA-backed Microchip processor have also become newer reference points; the NASA-backed processor has a 100-times throughput goal.

Starfire's listed 75 TOPS ceiling and dedicated NPU put Intel's pitch closer to on-orbit AI inference.

Tom's Hardware did not present Starfire as a flight-qualified replacement for those existing processors.

Radiation Characterisation Is Still In Process

Intel lists radiation data for total-dose exposure, latch-up and single-event effects as characterisation still in process, according to Tom's Hardware.

The report said the part is not radiation-qualified yet and that Intel notes the specifications remain subject to change.

Tom's Hardware said Intel's use of Intel 18A for CPU and NPU tiles and Intel 3 for the GPU resembles the node split used in Clearwater Forest.

The article said smaller transistors hold less charge per stored bit, which can make leading-edge silicon more exposed to radiation-induced bit flips.

Intel's domestic-manufacturing pitch sits alongside its Trusted Foundry status and Pentagon programmes.

Tom's Hardware reported that RAMP-C and SHIP include Intel 18A and advanced packaging work, and said 18A yields are not expected to reach industry-standard levels until 2027.

Intel has not disclosed completed radiation qualification, final specifications, production pricing, named mission customers or confirmed flight dates for the two listed Starfire SKUs.

Share this article
inXf

Related articles

More
Orbitworks Sets UAE Satellite Launch With Nvidia AI But No Named Western Buyer
Chips & Semiconductors

Orbitworks Sets UAE Satellite Launch With Nvidia AI But No Named Western Buyer

Abu Dhabi-based Orbitworks plans to ship the UAE-built Altair satellite to California this autumn, with Nvidia-powered on-board image processing and a 10-satellite target, but it has not named its first European customer.

Taiwan Visit Highlights How AI Demand Is Pulling Chip Power, Capacity and Pricing Into One Orbit
Chips & Semiconductors

Taiwan Visit Highlights How AI Demand Is Pulling Chip Power, Capacity and Pricing Into One Orbit

Jensen Huang’s Taiwan visit was centered on partner meetings, a planned discussion with TSMC Chairman C.C. Wei, and Nvidia appearances tied to the Computex cycle. The episode points to a broader market phase in which AI chip demand is increasingly shaped by foundry access, packaging readiness, memory costs and ecosystem coordination. Investors, suppliers and device makers should watch whether Taiwan-related expansion signals translate into concrete capacity, site, and pricing outcomes.

Anthropic’s Conway Points Claude Toward Always-On AI Agents
AI

Anthropic’s Conway Points Claude Toward Always-On AI Agents

Anthropic is preparing a Claude expansion that includes Conway, Orbit, Operon, memory upgrades and multilingual voice mode. The move signals a shift from chat-based AI toward persistent assistants that can connect with external services and manage workspaces. Enterprises, developers and research teams could be affected if Claude becomes a broader agent platform.

Indosat Bases AI Grid Plan On 80% Mid-Band 5G Rollout
Telco & Connectivity

Indosat Bases AI Grid Plan On 80% Mid-Band 5G Rollout

Indosat says its Nokia 5G modernisation will add low-band 5G nationwide and mid-band 5G to about 80% of its network over three and a half years, The public record still lacks site counts, capex or named AI Grid customers.

Nokia-Indosat AI-RAN Deal Turns Indonesia Into A Live 5G Testbed
Telco & Connectivity

Nokia-Indosat AI-RAN Deal Turns Indonesia Into A Live 5G Testbed

Nokia and Indosat have moved their Nvidia-backed AI-RAN collaboration from demonstrations toward a nationwide 5G modernization plan, with live-network field trials targeted by the end of 2026 and broader commercial deployment expected in 2027.

Corning’s Fiber Deals Put AI Data Centers On A Connectivity Clock
Telco & Connectivity

Corning’s Fiber Deals Put AI Data Centers On A Connectivity Clock

Corning’s optical-fiber deals with Amazon, Meta and Nvidia show how AI data-center growth is turning connectivity supply into a strategic capacity issue for hyperscalers.

Keep Reading

More Stories

Latest
PixVerse Raises $439 Million As Video AI Valuation Tops $2 BillionAIJul 14, 2026PixVerse Raises $439 Million As Video AI Valuation Tops $2 BillionTechCrunch reported that Singapore-based PixVerse has raised a total of $439 million in a Series C extension and crossed a $2 billion valuation. The video-generation startup lists more than 150 million registered users, but has not named paying-user totals, enterprise revenue or independent benchmark results.Intel Commits 5 Billion Euros To Irish Fab 34 ExpansionChips & SemiconductorsJul 14, 2026Intel Commits 5 Billion Euros To Irish Fab 34 ExpansionA 5 billion euro Leixlip investment puts Fab 34 back at the centre of Intel's European manufacturing plan for AI and high-performance computing processors. Most of the spending is due by the end of 2027, while a large external foundry customer for the Irish capacity remains unnamed, according to TNW.Nadella Warns Enterprises Pay Twice For Proprietary AI ModelsAIJul 14, 2026Nadella Warns Enterprises Pay Twice For Proprietary AI ModelsMicrosoft CEO Satya Nadella warned companies using proprietary AI models may pay with both tokens and internal knowledge, according to TechCrunch. Solo.io, Vercel and OpenRouter supplied evidence of interest in on-premise and open models, while customer-level cost savings and security outcomes remain undisclosed.BrainCo Uses AI Decoding In Non-Invasive BCI Push After $280 Million RoundAIJul 14, 2026BrainCo Uses AI Decoding In Non-Invasive BCI Push After $280 Million RoundThe Hangzhou BCI company raised 2 billion yuan, or $280 million, while building non-invasive brain-computer-interface devices that use AI decoding, according to Indian Express. Public records still lack current revenue, shipment volumes, clinical trial results, product pricing and independent validation of the AI decoding accuracy.Anthropic Adds India Rupee Pricing But Leaves UPI Out Of Claude PaymentsAIJul 14, 2026Anthropic Adds India Rupee Pricing But Leaves UPI Out Of Claude PaymentsTechCrunch said Anthropic has begun showing Claude prices in Indian rupees for some users, with India accounting for 5.8% of global Claude usage. The rollout covers local currency display but still relies on cards or app-store billing rather than UPI, while Mythos 5 access remains limited outside the U.S.Maybank Sees AI Demand Driving Singapore Semiconductor GrowthChips & SemiconductorsJul 13, 2026Maybank Sees AI Demand Driving Singapore Semiconductor GrowthMaybank Investment Bank said AI-related demand should remain the main growth driver for Singapore semiconductor companies, while warning that large AI infrastructure spending still needs stronger revenue proof.Xi Jinping To Open Shanghai AI Governance Summit For First TimeCapital & PolicyJul 13, 2026Xi Jinping To Open Shanghai AI Governance Summit For First TimeXi Jinping will keynote China’s 2026 World Artificial Intelligence Conference from 17 to 20 July alongside a global AI governance meeting, TNW reported. The preview does not name member countries, confirm a charter, or identify which delegations would support Beijing’s proposed AI cooperation body.Cloudflare Sets September 15 Permission Rules For AI Agent CrawlersCapital & PolicyJul 13, 2026Cloudflare Sets September 15 Permission Rules For AI Agent CrawlersCloudflare will block Agent and Training crawler traffic by default on ad-supported pages from September 15 for several customer groups, while Search remains allowed, according to AI News. Enforcement still depends on crawler behaviour labels.Railway Raises $100 Million As AI Coding Pushes Cloud Deployment ClaimsCloud & Data CentersJul 13, 2026Railway Raises $100 Million As AI Coding Pushes Cloud Deployment ClaimsRailway raised $100 million in a Series B round led by TQ Ventures, with the cloud startup citing more than 10 million deployments each month and two million developers, VentureBeat reported. Railway described sub-second deployment, customer cost-saving claims and its own data-centre buildout, but public detail did not include audited benchmarks, full enterprise contract values or customer-by-customer deployment scope.AICB Survey Says 44% Of Malaysian Banks Are Developing AI ReadinessFintech & Digital PaymentsJul 13, 2026AICB Survey Says 44% Of Malaysian Banks Are Developing AI ReadinessTechNode Global, citing source-owned material from AICB, reported that AICB’s Malaysian banks AI readiness survey put 44% of banks and development finance institutions in the developing stage of AI readiness. AICB said the survey drew on responses from 87 senior leaders, but the public source record did not name individual banks, provide production deployment counts, publish institution-level budgets or customer-impact metrics, or set out a timetable for closing the governance gaps.Deutsche Telekom Says ChatGPT Tools Reach 50,000 Monthly UsersAIJul 13, 2026Deutsche Telekom Says ChatGPT Tools Reach 50,000 Monthly UsersOpenAI's own customer case study says more than 50,000 monthly active users now use ChatGPT and API tooling inside Deutsche Telekom, with AI tool usage up 546% since the start of 2026. The source lists employee, customer-care, voice and network-operation uses; business outcome and performance claims are not independently verified in the public record.OpenAI Lists Family Product Role As ChatGPT Parent Use RisesAIJul 13, 2026OpenAI Lists Family Product Role As ChatGPT Parent Use RisesOpenAI is recruiting a product manager for family, caregiver and older-adult experiences as Sensor Tower estimates cited by TechCrunch showed ChatGPT's 35-and-older user share rising to 31% in Q2 from 26% a year earlier.