For most of January 2025 DeepSeek was a name known to people who read model cards. By the end of the month it had been the subject of a letter from the Irish Data Protection Commission, a block in Italy and a record one-day loss on the world's most valuable chipmaker. This briefing is an attempt to separate the three things that happened, which were a model release, a market reaction and a regulatory reaction, and to say what each of them meant. It is written at the start of February with the sources available then.
What was released
On 26 December 2024 DeepSeek, a Hangzhou lab spun out of the quantitative hedge fund High-Flyer, published DeepSeek-V3: a mixture-of-experts model with 671 billion parameters in total, 37 billion of them active for any given token, trained on 14.8 trillion tokens using 2.788 million hours of Nvidia H800 time.[1] The technical report priced that at $5.576m at an assumed $2 per GPU-hour, and said in the same passage that the figure covered the final training run only, not the research and ablation experiments that preceded it.[1]
On 20 January it released DeepSeek-R1, a reasoning model built on V3, under an MIT licence, alongside six smaller distilled models from 1.5 to 70 billion parameters.[2] The accompanying paper reported R1 on a par with OpenAI's o1 across maths, coding and science benchmarks: 79.8% on AIME 2024, 97.3% on MATH-500, 71.5% on GPQA Diamond and 49.2% on SWE-bench Verified.[3] The API price was $0.55 per million input tokens and $2.19 per million output tokens; o1 listed at $15 and $60.[2]
The claim that mattered to researchers was methodological. An intermediate model, R1-Zero, was trained by reinforcement learning against answers that could be checked automatically, with no supervised examples of humans reasoning, and developed behaviours such as re-checking its own work without being shown them.[3] R1 added a small amount of hand-written starting data and further training to make the output readable. Underneath, V3 combined three efficiency techniques that were each known but had not been stacked at this scale: the sparse expert layout, a compressed form of attention that cuts the memory needed to hold a long context, and training in 8-bit floating point rather than 16.[1]
The week the market noticed
- $589bn
- Fall in Nvidia's market value on Monday 27 January, the largest single-day loss in US market history [4]
- 17%
- Nvidia's one-day share price fall, its biggest since March 2020 [4]
- $0.55
- R1's price per million input tokens, against $15 for OpenAI's o1 [2]
Over the weekend of 25 and 26 January DeepSeek's free assistant reached the top of Apple's app chart in the United States, and the venture capitalist Marc Andreessen called R1 "AI's Sputnik moment".[5][6] On the Monday, Nvidia fell 17% and lost $589bn of market value, the largest one-day loss for any company on record; the Nasdaq 100 fell 3% and the S&P 500 1.5%, with ASML, Oracle, Broadcom and GE Vernova among the fallers.[4] The logic of the sell-off was simple. If frontier models could be trained for millions rather than billions, demand for chips and for the power to run them would be lower than the market had priced.
Regulators moved almost as fast. On 28 January Italy's data protection authority asked DeepSeek what personal data it collected, where it was stored and on what legal basis. On 29 January the Irish Data Protection Commission wrote to the company requesting information on the data processing conducted in relation to data subjects in Ireland; because DeepSeek has no European establishment in Ireland, the DPC is not its lead supervisory authority in the way it is for the American platforms.[7] On 30 January the Italian regulator, having called DeepSeek's reply "completely insufficient", blocked the app for Italian users and opened an investigation. DeepSeek's position was that it did not operate in Italy and that European law did not apply to it.[8] The same day OpenAI told journalists it had evidence DeepSeek had used its models' outputs to train its own, a practice known as distillation that OpenAI's terms prohibit.[9]
What the $6m figure means
The number that carried the story was the wrong one to argue about, and by the end of the week the better analysts had said so. SemiAnalysis, a semiconductor research firm, estimated on 31 January that DeepSeek had access to around 50,000 Nvidia Hopper-generation GPUs, that its server capital expenditure was about $1.6bn, and that the $6m covered the GPU rental for one pre-training run and nothing else: no research, no failed experiments, no data, no salaries.[10] That is also what DeepSeek's own report said, in a sentence most coverage did not reach.[1]
Anthropic's chief executive Dario Amodei published a long essay on 29 January making the other half of the argument. Algorithmic progress, he wrote, shifts the cost of a given capability by roughly four times a year, so a model matching American models seven to ten months old at a lower cost was to be expected. "DeepSeek-V3 is not a unique breakthrough," he wrote, but "an expected point on an ongoing cost reduction curve."[11] His concern was not the model. It was that a Chinese lab had reached the point first, and what that implied for export controls.
Both readings are right and they are not in tension. The training run was cheap and the organisation was not. The trend line said this was coming and the surprise was who published it, how openly, and with what price tag. A business does not need to adjudicate the geopolitics to draw the practical conclusion: the cost of a given level of capability was falling faster than most procurement cycles, and would keep doing so.
What it changed
- Open weights reached the reasoning frontier. Before R1 the best open models trailed the best closed ones by something like a year on capability. R1 put an MIT-licensed model within reach of o1, and its distilled versions put usable reasoning on a single consumer GPU.[2][3]
- The price anchor moved. OpenAI released o3-mini on 31 January at $1.10 per million input tokens and $4.40 per million output tokens, a fraction of o1's price, and every lab's pricing conversation changed with it.[12]
- The recipe was public. Reinforcement learning against verifiable answers, with the specific training method DeepSeek described, was reproduced by academic groups within weeks and became the default way to build a reasoning model.[3]
- Where the data goes became a board question. The regulatory letters were about the consumer app and the API hosted in China, not about the weights. The distinction matters, and most of the coverage missed it.[7][8]
What it did not change
The closed labs were still ahead, and stayed ahead. o3-mini arrived within eleven days of R1. The full R1 model needs a rack of data-centre GPUs to serve at a useful speed, and the distilled versions that run on a laptop are noticeably weaker than the headline numbers suggest. A reasoning model is still a language model: it produces confident wrong answers, it is not a source of record, and it is no safer to paste client data into than any other service, wherever the servers are. And for the vast majority of Irish businesses the question in January 2025 was not which frontier model to bet on. It was which of their own processes were worth putting any model on at all, and that question was unchanged by the week's events.
