{"id":230633,"date":"2026-01-10T09:08:06","date_gmt":"2026-01-10T09:08:06","guid":{"rendered":"https:\/\/www.newsbeep.com\/il\/230633\/"},"modified":"2026-01-10T09:08:06","modified_gmt":"2026-01-10T09:08:06","slug":"how-do-you-value-a-company-like-nvidia","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/il\/230633\/","title":{"rendered":"How do you value a company like Nvidia?"},"content":{"rendered":"<p>The new year has started in much the same way as the last one ended: with a lot of talk about bubbly valuations. The fact the S&amp;P 500 hit an all-time high this week,\u00a0didn\u2019t help.<\/p>\n<p>But how do you value what is undoubtedly the stock of the decade?<\/p>\n<p>Nvidia, the world\u2019s most valuable company, has seen its share price rise over 3,000 per cent since the start of 2020. It\u2019s now worth $4.5tn.\u00a0<\/p>\n<p>But is it really?<\/p>\n<p>The company has risen from being the leading maker of semiconductors for gaming consoles to being the leading processor for bitcoin mining and, now, the poster child of AI.<\/p>\n<p>These successes spring from the decision by founder and chief executive <a href=\"https:\/\/www.ft.com\/content\/11a018f4-95e0-41c2-99d8-aff105328a0b\" title=\"\" data-trackable=\"link\" rel=\"nofollow noopener\" target=\"_blank\">Jensen Huang<\/a> many years ago to use parallel processing of data, which made Nvidia\u2019s chips faster and more energy efficient than its rivals.\u00a0<\/p>\n<p class=\"n-content-recommended__title o3-type-body-highlight\">Recommended<\/p>\n<p><a href=\"https:\/\/www.ft.com\/content\/ae4f28a0-39f9-48e6-bdf4-09321b03c236\" data-trackable=\"image-link\" data-trackable-context-story-link=\"image-link\" tabindex=\"-1\" aria-hidden=\"true\" rel=\"nofollow noopener\" target=\"_blank\"><img decoding=\"async\" class=\"o-teaser__image\" src=\"https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fimages.ft.com%2Fv3%2Fimage%2Fraw%2Fhttps%253A%252F%252Fd1e00ek4ebabms.cloudfront.net%252Fproduction%252F2b098f34-7832-4bff-bbed-9e258593d217.jpg%3Fsource%3Dnext-article%26fit%3Dscale-down%26quality%3Dhighest%26width%3D700%26dpr%3D1?source=next&amp;fit=scale-down&amp;dpr=2&amp;width=240\" alt=\"Jensen Huang gestures while speaking, seated on stage with computer hardware components displayed nearby.\"\/><\/a><\/p>\n<p>More recently that has made them the first choice for AI \u201ctraining\u201d \u2014 giving Nvidia power to raise prices and lifting operating margins from the mid-teens a decade ago to about 60 per cent last year.\u00a0<\/p>\n<p>The company should make an operating profit of more than $120bn this year \u2014 more than the total value of all but the four biggest FTSE 100 companies. Its recent <a href=\"https:\/\/www.ft.com\/content\/3584197e-a99a-4a06-9386-dc65cf603f45\" title=\"\" data-trackable=\"link\" rel=\"nofollow noopener\" target=\"_blank\">$20bn deal<\/a> with AI chip start-up Groq is expected to help maintain its position in the high specification part of the market. <\/p>\n<p>But are its shares now too expensive? (Spoiler alert: we don\u2019t own Nvidia.)<\/p>\n<p>The first consideration is the competition. It may help to think of Nvidia chips as high-end sports cars. Great if you want huge power. But it\u2019s dangerous to assume AI\u2019s future is dependent on all that roar and thrust.\u00a0<\/p>\n<p>When the Chinese DeepSeek AI engine appeared last year, Nvidia\u2019s shares sold off sharply, as DeepSeek\u2019s results seemed comparable with those of OpenAI (Nvidia\u2019s largest client) and apparently involved fewer expensive, high-end chips.\u00a0DeepSeek has started 2026 with a research paper offering a more efficient way of training large language models, which may further reduce reliance on the most powerful chips.\u00a0<\/p>\n<p>Alphabet has developed its own Tensor chips (think decent saloon car). Only a couple of months ago it launched its Gemini 3 AI model, powered\u00a0by these. Elsewhere, Anthropic AI relies largely on chips designed by Alphabet and Amazon (AWS). <\/p>\n<p>The Gemini 3 and Anthropic models both outscore OpenAI\u2019s ChatGPT in some criteria. This helps explain the 60 per cent rise in Alphabet\u2019s share price last year.\u00a0(I confess that we missed much of this, wrongly believing Alphabet would be challenged by AI and the Department of Justice probe. We have remedied this now.)\u00a0<\/p>\n<p>In valuing Nvidia you can\u2019t ignore OpenAI. Nvidia intends to supply its client\u2019s data centres with over $100bn of Nvidia chips. It\u2019s too cynical to regard this arrangement as part of a \u201cPonzi scheme\u201d, which some have done \u2014 it\u2019s sensible to back important commercial partners. But OpenAI isn\u2019t expected to make an actual profit this decade, so we count this element of Nvidia\u2019s future income as part revenue, part loan.\u00a0<\/p>\n<p>Next up is the energy issue. The current \u201ctraining\u201d phase for AI chips involves them learning patterns in huge datasets and uses massive amounts of electricity. And there\u2019s the rub. Power limitations could pull the plug on the heady revenue growth numbers underpinning many AI stock forecasts. There are no easy solutions. Build an adjacent gas turbine? Sorry, manufacturers have sold out even before you apply for planning permission. Some data centre companies have begun repurposing jet engines, but this looks an expensive workaround.\u00a0<\/p>\n<p>If Anthropic floats on the Nasdaq this year it will be valued on hoped-for future cash flows. Power constraints could undermine sales and hit those valuations substantially. It\u2019s a good illustration of how a stock market boom can peter out when exposed to the dull light of reality.<\/p>\n<p>Valuing semiconductor stocks is always tricky. An old mate of mine compares it with valuing cement companies \u2014 both industries are capital-intensive and cyclical. These are not buy-and-hold stocks. Buy them at the bottom of the cycle and sell at the top.\u00a0<\/p>\n<p>Perhaps counter-intuitively, when their earnings have plunged, their share price\/earnings ratio may look ridiculously high for a while \u2014 that\u2019s when you buy. Similarly, low-looking price\/earnings ratios come at the cycle peak, so you sell when price\/earnings ratios look low.<\/p>\n<p>Nvidia outsources the manufacturing of its designs, thereby avoiding the eye-watering capital costs of building fabrication plants. So it\u2019s less capital-intensive than it might be. And AI should keep chip demand \u2014 and therefore sales \u2014 growing for a while. Huang\u2019s speech in Las Vegas this week focusing on the power of Nvidia\u2019s latest chips and its role in robotics may have reassured some investors that sales growth will continue, but if margins have peaked that could have little impact on profits.\u00a0\u00a0<\/p>\n<p>To my mind, the AI boom is moving to its next phase, which could benefit other AI-related stocks we own, such as Broadcom and Taiwan Semiconductor. These have been doing nicely. We also have sizeable holdings in Alibaba and Baidu, Beijing\u2019s AI champions.<\/p>\n<p>Apple, which we bought a few months ago, appears to be leaving the big spenders to develop the cutting-edge stuff, using more third-party AI technology within its products. This hybrid strategy could prove smart.\u00a0<\/p>\n<p>Some so-called \u201cAI losers\u201d could turn a corner in 2026. Software companies, such as holdings Salesforce\u00a0and SAP, have been seen as threatened by AI, but it could instead enable them to enhance significantly what they offer.<\/p>\n<p>The year ahead will be interesting. I\u2019ll undoubtedly have got some of this wrong \u2014 I wasn\u2019t taught divination at school. But over the years I\u2019ve found buying sensible stocks at a sensible price and diversifying is a reliable investment strategy. I like Nvidia, but I don\u2019t think it\u2019s a sensible price. And I think there are better alternatives from an investment perspective.<\/p>\n<p>Simon Edelsten is a fund manager at Goshawk Asset Management<\/p>\n","protected":false},"excerpt":{"rendered":"The new year has started in much the same way as the last one ended: with a lot&hellip;\n","protected":false},"author":2,"featured_media":230634,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[345,343,344,85,46,125],"class_list":{"0":"post-230633","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-artificial-intelligence","8":"tag-ai","9":"tag-artificial-intelligence","10":"tag-artificialintelligence","11":"tag-il","12":"tag-israel","13":"tag-technology"},"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/posts\/230633","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/comments?post=230633"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/posts\/230633\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/media\/230634"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/media?parent=230633"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/categories?post=230633"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/tags?post=230633"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}