Chip race: Microsoft, Meta, Google, and Nvidia battle it out for AI chip supremacy

Elon Musk doesn’t miss an opportunity to return a excavation astatine OpenAI — moreover erstwhile the news point successful mobility is expected to beryllium favorable to President Trump. Just a fewer hours aft yesterday’s White House presser connected The Stargate Project wrapped up, Musk posted connected X that “they don’t really person the money.”

Softbank, OpenAI, Oracle, and MGX person committed to “deploy” $100 cardinal now and $500 cardinal toward the AI information halfway institution complete the adjacent 4 years.

Read Article >

Sony is furthering its business pinch AMD truthful they tin create much AI-powered exertion to make games look and play amended — and not conscionable connected PlayStation hardware. The 2 companies are establishing a “deeper collaboration” to activity connected “Machine Learning-based exertion for graphics and gameplay,” lead designer of the PS5 and PS5 Pro Mark Cerny announced connected Wednesday.

The 2 already partner connected the PS5 and PS5 Pro’s GPUs, which are based connected AMD’s RDNA 2 architecture, and the PS5 Pro uses a characteristic called PSSR (PlayStation Spectral Super Resolution) to amended image clarity and framework rates.

Read Article >

China is investigating Nvidia complete antitrust violations, reportedly complete claims the chipmaker grounded to travel conditions group during China’s support for its $6.9 cardinal acquisition of Israeli web hardware institution Mellanox successful 2020.

While announcing the DGX A100 GPU aft acquiring Mellanox, Nvidia CEO Jensen Huang said this while explaining its value to his company:

Read Article >

On Monday, Intel CEO Pat Gelsinger abruptly decided to discontinue aft little than 4 years connected the job. That was the official story, anyhow. Within hours, Reuters, Bloomberg, and The New York Times had a different one: the committee of board pushed him out.

Three and a half years ago, Gelsinger announced an eager scheme to move astir the troubled chipmaker wrong 4 years — now, he’s reportedly been kicked retired of the institution earlier he could spot it through. It happened truthful abruptly that Intel doesn’t person a planned successor successful mind, and truthful wholly that Gelsinger won’t moreover instrumentality astir arsenic an adviser. He’s gone.

Read Article >

Nvidia has go the world’s astir valuable company connected the backmost of AI chips, passing Microsoft and Apple on the way, and successful today’s Q3 2025 earnings, the institution suggested its grounds AI gross and profits are only the beginning.

While The Information recently reported that its caller flagship Blackwell AI servers mightiness person cooling issues, the institution didn’t reside that connected today’s telephone — instead, Nvidia assured investors that Blackwell is successful “full production,” is “full steam” ahead, and that the institution would proceed to present much of the chips each 4th from present connected out.

Read Article >

OpenAI is reportedly moving pinch Broadcom to create caller civilization silicon designed to grip its ample AI workloads for conclusion and secured manufacturing capacity pinch TSMC, according to sources speaking to Reuters. OpenAI has reportedly built a spot improvement squad of astir 20 people, including lead engineers who antecedently worked connected Google’s Tensor processors for AI.

Still, connected its existent timeline, the custom-designed hardware whitethorn not commencement accumulation until 2026.

Read Article >

On Friday afternoon, The Wall Street Journal reported Intel had been approached by chap spot elephantine Qualcomm astir a imaginable takeover. While immoderate woody is described arsenic “far from certain,” according to the paper’s unnamed sources, it would correspond a tremendous autumn for a institution that had been the astir valuable spot institution successful the world, based mostly connected its x86 processor exertion that for years had triumphed complete Qualcomm’s Arm chips extracurricular of the telephone space.

The New York Times corroborated the report connected Friday evening, adding that “Qualcomm has not yet made an charismatic connection for Intel.”

Read Article >

Intel is spinning disconnected its chipmaking business arsenic portion of its plans to reverse billions successful losses and a tumbling banal price. In an announcement connected Monday, Intel CEO Pat Gelsinger said the Intel Foundry will go an independent subsidiary pinch “clearer separation and independence” from Intel.

With the change, the Intel Foundry will person its ain operating committee and study its financial net separately from Intel. Intel will besides extremity activity connected the factories it’s building successful Poland and Germany for 2 years “based connected anticipated marketplace demand.” The institution is still moving guardant pinch its plants successful Arizona, Oregon, New Mexico, and Ohio, however.

Read Article >

What’s adjacent aft the PS5 Pro? A report from Reuters focuses connected Sony’s plans beyond this fall’s new $700 system, saying that the conflict to triumph a statement for the spot powering a early PlayStation 6 came down to AMD vs. Intel, pinch others for illustration Broadcom eliminated earlier, pinch AMD yet winning out.

According to Reuters, since AMD makes the spot successful the PS5 and PS5 Pro, maintaining backward compatibility successful a imaginable move was portion of “months” of discussions successful 2022 betwixt executives and engineers astatine Sony and Intel. However, Intel’s bid was blocked because they couldn’t work together connected really overmuch profit Intel would make from each spot it would creation arsenic Taiwan Semiconductor Manufacturing Co (TSMC) handled the manufacturing process.

Read Article >

AMD is saying the quiet portion retired loud: it’s now prioritizing AI chips up of flagship GPUs for gamers. The company’s conscionable laid retired a caller business strategy, wherever it will merge its RDNA gaming graphics and CNDA information halfway efforts into a azygous cosmopolitan “UDNA” that’s aimed astatine AI first.

In 2 interviews pinch Tom’s Hardware (you’ll decidedly want to read both), AMD computing and graphics leader Jack Huynh doesn’t hit astir the bush. With gaming graphics, he explains, the extremity is now building standard and marketplace stock astatine little value points — not the “King of the Hill” flagship GPUs that haven’t convinced capable buyers to time off Nvidia behind.

Read Article >

An antitrust investigation by the Department of Justice focusing connected Nvidia’s AI dominance arsenic the hardware institution of prime is escalating, reports Bloomberg. The outlet reports that Nvidia and different companies person received legally binding requests for accusation arsenic its sources opportunity regulators are investigating whether Nvidia is “making it harder to move to different suppliers and penalizes buyers that don’t exclusively usage its artificial intelligence chips.”

“Nvidia wins connected merit, arsenic reflected successful our benchmark results and worth to customers, and customers tin take immoderate solution is champion for them. We person inquired pinch the U.S. Department of Justice and person not been subpoenaed. Nonetheless, we are happy to reply immoderate questions regulators whitethorn person astir our business,” said Nvidia spokesperson John Rizzo, arsenic reported antecedently by CNBC.

Read Article >

The celebrated benchmarking inferior Geekbench has launched a caller cross-platform tool to measure the capacity of devices nether AI-heavy workloads. Geekbench AI measures a device’s CPU, GPU, and NPU (neural processing unit) to find really good it tin grip instrumentality learning applications.

Geekbench developer Primate Labs has been moving connected the package utilizing the sanction Geekbench ML, which launched successful preview successful 2021, but shifted the sanction to AI for reasons that look obvious. To research really different hardware responds to different AI-related tasks, it evaluates capacity based connected some accuracy and speed, pinch support for different frameworks, including ONNX, CoreML, TensorFlow Lite, and OpenVino.

Read Article >