At the start of 2024, Walmart teamed up with OpenAI to launch Instant Checkout – a generative AI designed for instant payment. The system frequently "locked up" on complex carts, required operator intervention, saw conversion drop to just 2% instead of the targeted 10‑15%, and processing times exceeded the promised five seconds. Six months later, the project was declared a failure and Walmart announced a strategic shift.
Walmart did not sever its partnership with OpenAI but abandoned its home‑grown "brain" in favor of established large language models – OpenAI's ChatGPT and Google's Gemini. Central to the new approach is the chat bot Sparky, originally built for internal product recommendations. It has now been embedded into the conversational interfaces of these LLMs, allowing shoppers to ask product questions without a separate app.
Switching to off‑the‑shelf models accelerated the rollout of new features: ChatGPT users are already accustomed to AI advice and can now interact with Sparky directly. Every query is monetized – providers charge per token, and a portion of that revenue is shared with Walmart as a licensing fee. Access to massive models gives Sparky context about trends, promotions and shopper preferences (with customer consent), boosting personalization and potentially raising average basket size.
However, the strategy deepens Walmart's technological reliance on external vendors. A price hike or API restriction from OpenAI or Google could sharply cut Walmart’s earnings. Data privacy concerns intensify: queries contain purchase details and addresses, travel through providers' infrastructure, and attract regulator scrutiny; a leak could damage reputation and trigger fines.
Competitors are strengthening their own ecosystems. OpenAI is developing its own e‑commerce tools, while Google is investing in Gemini‑based shopping assistants that can recommend rival products with comparable personalization. This challenges the exclusivity of Walmart’s access to its customer base.
For investors, the key takeaway is that moving from a proprietary Instant Checkout to Sparky integration aims to slash development costs and tap the scale of proven platforms, but it comes at the expense of control over the tech stack and heightened operational risk. The financial impact remains unclear, yet early quarterly reports show a 35% jump in Sparky queries after its launch inside ChatGPT, indicating monetization potential.
Strategically, companies should prepare a "Plan B": develop their own APIs for direct access to Sparky and invest in alternative models such as the open‑source LLaMA. This would reduce dependence on OpenAI and Google and provide greater flexibility around data confidentiality.
Why it matters: instant checkout did not become a competitive edge; ready‑made LLMs enable faster service rollout, but the price is loss of data control and vendor risk. CEOs need to weigh speed against platform dependency and plan backup technological solutions.