Recently, Intel CEO Pat Gelsinger has been in the headlines since he recently concluded the company’s launch of its next-generation CPUs for servers and clients. He had a conversation about the state of the semiconductor industry with MIT engineering students during a recent press trip. He caused a stir when he said that Nvidia is currently leading the AI space, despite Intel having the potential to do so—only because the latter company was lucky.

In answer to a professor who inquired about Intel’s efforts regarding AI hardware, Gelsinger made his remarks. In response to this question, Gelsinger recapped Intel’s disastrous experience with GPUs and “throughput computing” (as opposed to scalar computing), noting that when the corporation fired him 11 years prior, it also canceled its discrete GPU project, known as Larrabee. Gelsinger claims that if the business had continued with that initiative, Intel would currently be at the top of the AI sector. Rather, Nvidia finds itself in charge, which Gelsinger attributes to Nvidia CEO Jensen Huang’s “extraordinarily fortunate” situation.

He contends that Nvidia was always primarily concerned with graphics, which is true given that the company was founded on gaming before branching out into artificial intelligence and data centers. The corporation didn’t even want to sponsor its first large-scale AI effort, according to Gelsinger, because running the large data sets they were utilizing required a supercomputer. For fifteen years, Intel did nothing about “throughput computing,” whereas Nvidia gradually moved toward it. Upon his return to the firm in 2021, Gelsinger promptly rectified this error by resuming the project, which is currently recognized as its Arc graphics line of GPUs.

Here, Gelsinger isn’t entirely incorrect, as PC Gamer points out. Apart from the fact that Nvidia just makes the right products at the appropriate time, which is also true, his remarks also seem to be more of a jab at Intel’s lack of vision than anything about Nvidia. The website observes that while it appears plausible that Nvidia did not foresee the present AI frenzy, the company has persistently advanced the platform’s development. With what are widely regarded as the top goods in the industry—including its CUDA software and hardware—it has arrived at the ideal time. Is it pure chance, or is it a well-informed wager?

Businesses in the computing business, like everything else, are always placing bets on future events. Those wagers can occasionally pay off and occasionally go completely bust. Although Intel has had more failures than successes, Gelsinger hopes to “course correct” with the company’s current offering. In the field of artificial intelligence, Gelsinger has declared that the Gaudi 3 AI accelerator that would be released in 2024 will be able to compete with Nvidia’s H100 and AMD’s MI300.

The Moon has canceled one trip, but expects a spike in guests. The first lunar lander launch by Houston-based Intuitive Machines, scheduled for January, has stated that it is delaying the IM-1 mission until at least mid-February. The business cites modifications to SpaceX’s launch plan as evidence that the delay was inevitable.

Though Intuitive Machines took care to avoid pointing fingers, SpaceX is most likely to blame for this. This month’s scheduled launch of the US Military’s X-37B spacecraft by the business was postponed due to technical difficulties with the rocket. It is now anticipated that the X-37B won’t launch into space on the Falcon Heavy until at least December 28. Intuitive Machines will therefore need to wait until February.

There is considerable leeway in those dates, so why are we discussing an IM-1 launch in February? SpaceX uses Kennedy Space Center’s historic Launch Complex 39A to launch its Falcon 9 and Falcon Heavy rockets. But the rockets require different hardware, and changing everything out takes roughly three weeks. Even the smallest interval this year between the launches of the Falcon 9 and the Falcon Heavy was more than two weeks. The launchpad just cannot be prepared in time for the IM-1 mission window to close.

Although Intuitive Machines has stated that it hopes to launch in the middle of February, it has not provided a specific date. The company’s Nova C lander (shown above) will launch and go toward the Moon’s south polar zone, which India’s Chandrayaan-3 mission has partially explored. India outperformed Russia as the first country to land on this part of the Moon, following the crash of the Roscosmos Luna-25 mission this past summer. NASA payloads as part of the Commercial Lunar Payload Services (CLPS) program and private commercial experiments will also be carried on board Nova C.

There’s no assurance that Nova C will reach the lunar surface. Even though humans have sent to the Moon many years ago, landing on the lunar surface is still extremely challenging. It’s not just Roscosmos that has lost a lander. Israel’s SpaceIL lost the Beresheet lander in 2019, while Japan’s ispace also witnessed the collision of its Hakuto-R lander with the surface this year.

It appears that February will be a busy month for lunar operations. Additionally, Astrobiotic has an almost completed lunar lander. With both commercial and NASA payloads, its Peregrine lander is scheduled to launch in January on a United Launch Alliance Vulcan Centaur rocket and descend to the Moon on February 23.

Even though AMD’s new Instinct MI300 chips are still in the early stages of development, the firm has already announced that one of its supercomputers in Germany would use them. AMD has said that “Exascale Supercomputing Is Coming to Stuttgart” and that it will construct two machines: an exascale machine akin to its present Frontier supercomputer and one that will update an existing system to 39 PFLOPS. Hunter and Herder are the names of the two robots; the former will go online in 2025, while the latter is scheduled to debut in 2027.

The University of Stuttgart and Hewlett Packard Enterprise have inked a new deal that will result in the two new supercomputers. The group will use it to enhance its current Hawk supercomputer and set up a second machine at the Stuttgart-based HLRS, a research institute and supercomputer facility, in the future. This is the first supercomputer contract for AMD’s new MI300A technology, which integrates a CPU, GPU, and high-bandwidth memory into a single package. That’s the major news here. The center’s flagship supercomputer, Hawk, which has a speedy 26 PFLOPS, will be equipped with these data center “APUs”. This computer is neither outdated nor sluggish; in 2020, it debuted at position 16 on the Top 500 list. Having said that, there’s no malice intended; we completely comprehend the need to update a PC.

The center’s Hawk system is state-of-the-art for 2020; it currently runs on AMD Epyc “Rome” CPUs and Nvidia A100 GPUs. With 136 nodes, the upcoming Hunter version will be built on the HPE Cray EX4000 supercomputer. According to the organization, this new version represents a significant advancement for the center since it initiates a phased shift from largely depending on CPUs to more energy-efficient GPUs. Thanks to the integrated memory and APU design, updating the system to AMD’s new MI300A will increase efficiency by 80% and increase output to 39 PFLOPS. World-class industrial and academic research, high-performance data analytics (HPDA), simulations, and artificial intelligence (AI) will all be conducted on this computer.

While Hunter is anticipated to be operational by 2025, AMD accelerators will also be used in the construction of a new Exascale machine called Herder. Since this system won’t be turned on until 2027, the hardware will probably be whatever the next MI300 generation is, which will be decided in 2025. Additionally, AMD powers Frontier, the only Exascale supercomputer in the world, which, despite its June 2022 launch, continues to occupy the top spot on the podium at Top500.org. This gigantic system will use GPUs as its major computation source, which we think will become the standard in 2027, unlike Hawk and Hunter. The center claims that the advantages of GPU design’s huge parallelism over CPU design are the reason behind this shift.

Because of their comparable masses and sizes, astronomers recognized Venus as Earth’s sister planet; however, this was before they discovered Venus’s true differences. The surface of this planet is intensely hot and surrounded by a suffocating layer of sulfur, carbon dioxide, and water vapor. Scientists are very interested in how Venus came to be in this state since it is the victim of a runaway greenhouse effect. The ease with which Earth might devolve into a comparable runaway condition is demonstrated by a recent analysis from the University of Geneva (UNIGE).

The technical aspects of this research, the first to successfully replicate the full runaway greenhouse scenario, were handled by the Genevan team with assistance from the CNRS facilities in Bordeaux and Paris. The simulation started with a similar-looking but lovely exoplanet to Earth. Once the process has begun, the scientists discovered that sufficient warming makes it nearly impossible to reverse the changes in cloud cover and atmospheric structure—hence the term “runaway.”

One of the most common greenhouse gases on Earth, carbon dioxide, makes up the majority of Venus’ atmosphere. The concentration of water vapor, the researchers discovered, is what throws the balance in favor of a runaway greenhouse effect. Elevated concentrations of water vapor raise temperatures similar to those caused by carbon dioxide, which in turn promotes ocean evaporation and adds additional water vapor to the atmosphere. As a result, evaporation and temperature rise in a feedback cycle that continues until all of the water turns into gas.

This level of water vapor has a critical threshold, at which the planet can no longer cool. The temperature rises to several hundred degrees and the oceans completely evaporate after that, according to Guillaume Chaverot, the study’s author. Prior research on runaway greenhouse conditions concentrated on the runaway’s temperature condition or the planet’s post-process state, but this study’s simulation of the point of no return revealed fresh information.