Meta begins testing its first in-house AI training chip
The world's biggest social media company has begun a small deployment of the chip and plans to ramp up production for wide-scale use if the test goes well, the sources said.
The push to develop in-house chips is part of a long-term plan at Meta to bring down its mammoth infrastructure costs as the company places expensive bets on AI tools to drive growth.
Meta, which also owns Instagram and WhatsApp, has forecast total 2025 expenses of $114 billion to $119 billion, including up to $65 billion in capital expenditure largely driven by spending on AI infrastructure.
One of the sources said Meta's new training chip is a dedicated accelerator, meaning it is designed to handle only AI-specific tasks. This can make it more power-efficient than the integrated graphics processing units (GPUs) generally used for AI workloads.
Meta is working with Taiwan-based chip manufacturer TSMC to produce the chip, this person said.
The test deployment began after Meta finished its first "tape-out" of the chip, a significant marker of success in silicon development work that involves sending an initial design through a chip factory, the other source said.
A typical tape-out costs tens of millions of dollars and takes roughly three to six months to complete, with no guarantee the test will succeed. A failure would require Meta to diagnose the problem and repeat the tape-out step.
Meta and TSMC declined to comment.
The chip is the latest in the company's Meta Training and Inference Accelerator (MTIA) series. The program has had a wobbly start for years and at one point scrapped a chip at a similar phase of development.
However, Meta last year started using an MTIA chip to perform inference, or the process involved in running an AI system as users interact with it, for the recommendation systems that determine which content shows up on Facebook and Instagram news feeds.
Meta executives have said they want to start using their own chips by 2026 for training, or the compute-intensive process of feeding the AI system reams of data to "teach" it how to perform.
As with the inference chip, the goal for the training chip is to start with recommendation systems and later use it for generative AI products like chatbot Meta AI, the executives said.
"We're working on how would we do training for recommender systems and then eventually how do we think about training and inference for gen AI," Meta's Chief Product Officer Chris Cox said at the Morgan Stanley technology, media and telecom conference last week.
Cox described Meta's chip development efforts as "kind of a walk, crawl, run situation" so far, but said executives considered the first-generation inference chip for recommendations to be a "big success."
Meta previously pulled the plug on an in-house custom inference chip after it flopped in a small-scale test deployment similar to the one it is doing now for the training chip, instead reversing course and placing orders for billions of dollars worth of Nvidia GPUs in 2022.
The social media company has remained one of Nvidia's biggest customers since then, amassing an arsenal of GPUs to train its models, including for recommendations and ads systems and its Llama foundation model series. The units also perform inference for the more than 3 billion people who use its apps each day.
The value of those GPUs has been thrown into question this year as AI researchers increasingly express doubts about how much more progress can be made by continuing to "scale up" large language models by adding ever more data and computing power.
Those doubts were reinforced with the late-January launch of new low-cost models from Chinese startup DeepSeek, which optimize computational efficiency by relying more heavily on inference than most incumbent models.
In a DeepSeek-induced global rout in AI stocks, Nvidia shares lost as much as a fifth of their value at one point. They subsequently regained most of that ground, with investors wagering the company's chips will remain the industry standard for training and inference, although they have dropped again on broader trade concerns.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Filipino Times
13 hours ago
- Filipino Times
PBBM Urges Filipinos to Use eGovPH App, Says It's as Simple as Facebook
President Ferdinand R. Marcos Jr. has encouraged more Filipinos to use the eGovPH app, assuring that it is as easy to navigate as popular social media platforms like Facebook. In Episode 3 of the BBM Podcast aired Thursday, Marcos said he had instructed developers to make the app simple enough for non-tech-savvy users to complete government transactions conveniently. Citing a government study estimating that 89 million Filipinos use Facebook, the President emphasized that the same skills can be applied to using eGovPH. The app consolidates multiple government services into a single mobile platform, in line with the administration's push for improved public service delivery and digitalization. Marcos said the goal is to reduce physical visits to government offices, even for older citizens or those unfamiliar with computers. He added that expanding online transactions is vital for the Philippines to compete globally and conduct business efficiently. 'It's absolutely imperative now that children know how to use a computer, the internet, and to work online. It's not an option anymore,' the President said.
Middle East Eye
21 hours ago
- Middle East Eye
As AI drives a new nuclear era, US tech giants and China reshape global power
As of March 2025, Google, Meta and Microsoft, the three biggest tech companies in the United States, pledged to triple their investments in nuclear energy by 2050. These plans highlight the sector's efforts to address one of the major challenges posed by the rapid expansion of artificial intelligence: its immense energy demands. This renewed interest in nuclear energy marks a new phase in the technology's history. The development of nuclear power began in the late 1930s, as the US and the Soviet Union raced to build the atomic bomb. Shortly after the Second World War, the technology was redirected towards civilian and commercial use, leading companies to begin developing nuclear plants under state supervision. However, by the late 1970s, concerns over safety, the declining cost of other renewable sources such as solar and wind, and the global prevalence of oil led many countries - particularly democracies - to scale back nuclear development. New MEE newsletter: Jerusalem Dispatch Sign up to get the latest insights and analysis on Israel-Palestine, alongside Turkey Unpacked and other MEE newsletters China, by contrast, continued to invest. Over the following decades, it steadily expanded its nuclear infrastructure, positioning itself as a global leader just as the rest of the industry languished, with little development and declining appeal. Today, the convergence of China's state-led strategy and the private ambitions of western tech giants marks the beginning of a new nuclear era - one set to transform energy systems, accelerate the AI race, and reshape global power. Driving forces The revival of nuclear energy has been driven by a convergence of technological, economic and strategic factors. First, the astronomical growth in energy demand to power AI has rendered traditional sources increasingly inadequate. Tech companies have turned to nuclear power for its enhanced capacity and relatively low carbon footprint. Meta alone aims to generate 4 gigawatts (GW) of energy - enough to power up to three million US homes - a figure that illustrates the enormous energy needs of today's tech giants. Only the largest corporate players can afford the astronomical costs of nuclear energy - but the promise of AI-driven profits is drawing billions in investment Second, although nuclear energy is relatively inexpensive to operate, the initial capital costs are extraordinarily high and often exceed projected budgets. For example, generating 4 GW of power can cost around $40bn - more than the total energy budget of many countries. Only the largest corporate players can afford such investments. The promise of substantial AI-driven profits is one key incentive, spurring the rise of nuclear energy start-ups and drawing billions in venture capital. Third, since the early 2000s, the industry has seen major technological breakthroughs, including small modular reactors (SMRs), which are easier to transport and assemble, and microreactors, which require less human oversight. Meanwhile, Gen IV reactors can produce more energy through advanced processes. These innovations coincide with AI's increasing energy requirements - and, crucially, some have been made possible by AI itself, which enables large-scale simulations and reduces human interaction with reactors, thereby improving safety. Global stakes China, too, is advancing many of these innovations - but through a centralised, state-led framework for nuclear development that stands in sharp contrast to the market-driven approach of western tech firms. Still, a notable degree of public-private collaboration remains. Full state backing of this technology reduces regulatory hurdles and increases coherence in the entire project. In addition, China has focused on thorium reactors - operationalising its first in April 2025 - in place of uranium, to reduce dependence on western supplies. This was aided by China's early 2025 discovery of sufficient domestic thorium reserves to power its nuclear programme for decades. This unfolding nuclear era carries far-reaching global and political implications, beginning with the growing influence of tech companies in a space once dominated by governments. Their emergence as major players in nuclear development creates friction with US regulatory authorities, as commercial ambitions often run up against stringent licensing and safety protocols. While President Donald Trump has ramped up support for the industry, regulatory oversight remains a critical point of tension - especially given the risk that companies might prioritise speed and profit over caution, a pattern well-documented across the corporate world. Schrodinger's president: Trump's tariffs exist and don't - except on China John Rees Read More » These innovations also intensify the strategic rivalry between the US and China. With tariff wars having intensified since Trump came to power, tensions are already at fever pitch. While the Cold War arms race was measured in warheads, this new nuclear age will likely be defined by the number of reactors built - and the AI technologies they power. Nuclear energy is becoming another front in a widening struggle for global dominance. At the same time, both tech companies and China are now looking to export these innovations, particularly smaller nuclear reactors, to developing countries to help recover costs. While tech firms may face regulatory hurdles in doing so, China could take advantage of its vast trade networks under the Belt and Road Initiative to sell this technology more freely - potentially leading to major geopolitical shifts and further challenging US hegemony. Energy transition The rise of nuclear technology is also expected to have ripple effects across existing energy sources - particularly coal, natural gas and oil, in that order of magnitude. The energy shift underway could redraw political and economic power in the Middle East Coal, already in decline, is expected to lose what remains of its foothold. Gas and oil may follow, as carbon taxes and climate regulations intensify pressure for cleaner alternatives. Over time, this transition could shift the foundations of political and economic power, creating new dynamics in regions such as the Middle East, where many of today's energy giants are based. Nuclear technology, once seen as a Cold War relic, is experiencing a silent resurgence in innovation and funding - one that is set to transform not only energy systems, but the geopolitical landscape for decades to come. The views expressed in this article belong to the author and do not necessarily reflect the editorial policy of Middle East Eye.
Arabian Post
a day ago
- Arabian Post
Mira Developments Expands into Oman with Major Investment
Arabian Post Staff -Dubai Mira Developments, a UAE-based luxury real estate developer, has made a significant move into Oman with the acquisition of a one million square foot plot of land in the Dhofar Governorate. This strategic expansion positions the company to tap into the growing demand for luxury properties in Salalah, particularly as the region gains attention for its potential in high-end tourism and residential developments. The announcement came shortly after Mira Developments hosted a high-profile three-day mastermind retreat that brought together over 100 top-tier real estate professionals. The retreat, which included brokers, architects, and representatives from luxury brands, marked a pivotal moment in the company's ambitious expansion strategy. The event served as both a networking platform and a launchpad for the company's upcoming developments in Oman, including the highly anticipated Mira Coral Bay project. ADVERTISEMENT Mira Developments' decision to enter the Omani market underscores its broader vision to become a leader in luxury branded living across the Gulf Cooperation Council region. With a focus on integrating premium design, sustainability, and lifestyle excellence, the company aims to set new standards for luxury communities in the region. The project will not only contribute to the local economy but also elevate the standards of residential living in Salalah, which is emerging as a prime destination for affluent residents and tourists. The retreat, which was set against the picturesque backdrop of Salalah's unique natural landscape and the refreshing Khareef season, proved to be the perfect setting for brainstorming and collaboration. The event featured a series of workshops and strategic roundtable discussions that aimed to push the boundaries of innovation in real estate. Participants explored opportunities for cross-sector collaboration and exchanged insights on future trends in luxury living, from sustainable architecture to cutting-edge smart home technology. The Dhofar region, known for its lush green landscapes and pristine coastline, offers a distinctive appeal for luxury developments. Salalah, in particular, is poised to benefit from its increasing popularity as a year-round destination. The Khareef season, which attracts tourists with its cool climate and stunning natural beauty, provides an ideal environment for high-end real estate projects designed to offer both comfort and exclusivity. Mira Coral Bay, the flagship development, is expected to feature world-class amenities, including bespoke residences, upscale retail spaces, and recreational facilities, all designed to cater to the discerning tastes of luxury buyers. The development will incorporate sustainable practices, aligning with Mira Developments' commitment to environmental responsibility. The company has also indicated that its approach will prioritise creating harmonious, eco-friendly communities that blend seamlessly with the natural surroundings of Salalah. Real estate experts have noted that the luxury property market in Oman is ripe for development, particularly in regions like Dhofar. With its stunning landscapes and growing infrastructure, Salalah offers an attractive proposition for high-end investors looking for new opportunities outside of the more saturated markets in the UAE and Saudi Arabia. The entry of Mira Developments into the Omani market is seen as a significant endorsement of the potential of Salalah as a luxury destination, and the project is expected to attract both regional and international investors. While Mira Developments' focus is on luxury properties, its broader vision extends beyond just creating high-end residences. The company aims to shape the future of luxury living by fostering innovation and collaborating with leading experts across industries. Through its strategic expansions, Mira Developments is positioning itself as a key player in the future of real estate development in the GCC region.
